Add incremental chunked reading to CborReader

[muhammad-othman]

Implements the API approved in #129718, adding incremental reading mode where the caller owns I/O and supplies data in chunks:

public partial class CborReader
{
    public CborReader(ReadOnlyMemory<byte> data, CborReaderOptions? options, bool isFinalBlock);
    public void SlideData(ReadOnlyMemory<byte> data, bool isFinalBlock);
    public void Reset(ReadOnlyMemory<byte> data, bool isFinalBlock);
    public bool TrySkipValue(bool disableConformanceModeChecks = false);
    public bool TrySkipToParent(bool disableConformanceModeChecks = false);
}

public partial enum CborReaderState
{
    NeedsMoreData,
}

• In non-final mode, PeekState() returns NeedsMoreData when the next token is incomplete in the buffer; any other state guarantees the single-token read won't fail on end of data. Multi-token reads (indefinite-length strings, ReadEncodedValue, semantic tag readers) still throw on truncation and restore the reader state, so the caller can slide in more data and retry.
• SlideData replaces the buffer and preserves nesting state; the caller must keep the unread bytes (BytesRemaining) at the start of the new buffer. Only the length is validated.
• Non-final mode is Lax-only in this version (per the proposal): other conformance modes track map keys by buffer offset, which cannot survive a slide. The validation is centralized so a future all-modes implementation has one place to rebase that bookkeeping.
• Final-block readers (all existing constructors and Reset(data)) are behaviorally unchanged.

Notes for reviewers:

• In non-final mode, reading at a root-value boundary throws CborContentException rather than InvalidOperationException, since the reader can't know the sequence ended until the final block arrives.
• Found but not fixed here: a root sequence ending in a dangling tag reports Finished, though it isn't well-formed. This predates this change; incremental readers deliberately match it for consistency. I think we should track it separately and will create an issue for it.

[azure-pipelines]

Azure Pipelines:
Successfully started running 3 pipeline(s).
13 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[dotnet-policy-service]

Tagging subscribers to this area: @dotnet/area-system-formats-cbor, @bartonjs, @vcsjones
See info in area-owners.md if you want to be subscribed.

[muhammad-othman]

@dotnet-policy-service agree company="Amazon"

[Copilot]

🔵 Needs a closer look

It adds new public API surface and changes core reader behavior, which warrants final maintainer review despite strong accompanying test coverage.

Pull request overview

This PR extends System.Formats.Cbor.CborReader with an incremental (chunked) reading mode where callers can supply data in multiple buffers, while preserving nesting state between refills.

Changes:

• Adds new public APIs to support incremental reading (isFinalBlock constructor/Reset, SlideData, TrySkipValue/TrySkipToParent) and introduces CborReaderState.NeedsMoreData.
• Updates PeekState() so non-final readers return NeedsMoreData for truncated next-token scenarios (instead of throwing), while multi-token semantic tag readers continue to throw on truncation and roll back state.
• Adds/updates tests to validate SlideData semantics, NeedsMoreData gating behavior, and resumable skipping across all split points.

File summaries

File	Description
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.cs	Adds isFinalBlock plumbing, SlideData, Reset overload, and checkpoint/state support needed for incremental mode.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.PeekState.cs	Implements NeedsMoreData behavior for non-final readers and token-availability gating.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.SkipValue.cs	Introduces TrySkipValue/TrySkipToParent and reworks skipping to return false for needs-more-data in non-final mode.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.Integer.cs	Adds DecodeCollectionLength to relax definite-length collection header handling for non-final buffers.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.Array.cs	Uses DecodeCollectionLength so array headers can be read even when contents arrive later.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.Map.cs	Uses DecodeCollectionLength and adjusts definite-length map size validation for non-final mode.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.Tag.cs	Ensures semantic tag readers treat NeedsMoreData as truncation (throw) to preserve atomic multi-token semantics.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReader.String.cs	Documents that returned memory views can be invalidated if the caller reuses buffers via SlideData.
src/libraries/System.Formats.Cbor/src/System/Formats/Cbor/Reader/CborReaderState.cs	Adds CborReaderState.NeedsMoreData with API docs.
src/libraries/System.Formats.Cbor/src/Resources/Strings.resx	Adds localized strings for new incremental-mode validation and exceptions.
src/libraries/System.Formats.Cbor/ref/System.Formats.Cbor.cs	Updates public surface area to include the new APIs and enum member.
src/libraries/System.Formats.Cbor/tests/System.Formats.Cbor.Tests.csproj	Includes the new test file in the test build.
src/libraries/System.Formats.Cbor/tests/Reader/CborReaderTests.SlideData.cs	Adds extensive incremental-mode tests (sliding, gating, truncation behavior, resumability).
src/libraries/System.Formats.Cbor/tests/Reader/CborReaderTests.SkipValue.cs	Adds tests for TrySkipValue/TrySkipToParent behavior in non-final mode.
src/libraries/System.Formats.Cbor/tests/Reader/CborReaderTests.Helpers.cs	Expands test vectors and splits invalid inputs into truncated-vs-malformed sets for incremental semantics.

Review details

• Files reviewed: 15/15 changed files
• Comments generated: 0
• Review effort level: Lite

[bartonjs]

Generally LGTM, but a few notes.

[Copilot]

Copilot review overview

🔵 Needs a closer look

The changes span a broad API and state-management surface requiring final human review.

Review effort: Lite
Findings: None

[Copilot]

Copilot review overview

🔵 Needs a closer look

A moderate retry-allocation issue remains unresolved, requiring changes and final human review.

Review effort: Lite
Findings: None

[muhammad-othman]

Thank you for your review, I've applied your suggestions to the PR, please let me know if there are anything that I missed.

[Copilot]

Copilot review overview

🟡 Changes recommended

Public state documentation and exception-contract wording need alignment before approval.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• Clarify NeedsMoreData after complete root-level values · New
• Broaden NeedsMoreData documentation beyond incomplete tokens · New

[am11]

@artl93, was the servicing label meant for this PR? It's not a backport. Looks like a mistake.

[Copilot]

Copilot review overview

🟡 Changes recommended

Correct the NeedsMoreData documentation and malformed XML documentation before approval.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• Fix malformed remarks XML paragraph tags · New
• Broaden NeedsMoreData documentation beyond incomplete tokens

Resolved since last review (1)

• Clarify NeedsMoreData after complete root-level values

[artl93]

@artl93, was the servicing label meant for this PR? It's not a backport. Looks like a mistake.

@am11 - you are correct. This still needs to get this backported to 11.0 to make the cutoff.

[Copilot]

Copilot review overview

🔵 Needs a closer look

Clarify the NeedsMoreData root-boundary documentation and resolve the exception-contract discrepancy.

Review effort: Lite
Findings: 1

Open (1)

• Broaden NeedsMoreData documentation beyond incomplete tokens

Resolved since last review (1)

• Fix malformed remarks XML paragraph tags

[bartonjs]

@muhammad-othman No more commits, please (and please, never force-push to an already open PR).

Every time you commit again you reset the CI clock, which is measured in hours.

[muhammad-othman]

@muhammad-othman No more commits, please (and please, never force-push to an already open PR).

Every time you commit again you reset the CI clock, which is measured in hours.

Sorry I was trying to get the commit history in good shape 😓 , no more commits from my side.

[bartonjs]

@muhammad-othman We squash-merge, so all of the individual commits more or less go away at merge time.

[Copilot]

Copilot review overview

🟡 Changes recommended

Resolve the public overload ambiguity and address the documented skip and allocation issues.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• New overload makes existing default-literal calls ambiguous · New

Resolved since last review (1)

• Broaden NeedsMoreData documentation beyond incomplete tokens

[bartonjs]

/ba-g Tests have already succeeded on multiple platforms, no platform-specific dependencies

[bartonjs]

/backport to release/11.0

[github-actions]

Started backporting to release/11.0 (link to workflow run)