feat(tests): cover a null EIP-8070 custodyColumns and repin to the current EIP blob

[spencer-tb]

Description

Add test_fcu_custody_columns_null, pinning the 2026-08-12 EIP-8070 clarification that a null custodyColumns on engine_forkchoiceUpdatedV4 is a blobpool no-op accepted with VALID, both on a client with no custody set and after one was configured, and test_fcu_custody_columns_identical for the other no-op case in the same rule, resending the current set. The blobs test now takes custody_columns_updates, an ordered list of custodyColumns values sent one forkchoice update each before getBlobsV4, where None sends an explicit null; leaving it unset sends no update, and an empty list is rejected by a validator with a unit test. Repin the suite to the current blob, which also carries the RLP empty-list clarification for elided blobs; the framework already encodes that. Execute-only suite, so check_eip_versions never runs its version item; the collection shows 9 cases and just static is clean.

These arms pin what the Engine API can observe: the value is well-formed, the forkchoice update completes with VALID, and getBlobsV4 still serves every cell afterwards. The resulting custody set only steers devp2p sampling of peer-announced transactions and is not observable here; that needs a devp2p eth/72 harness, which hive does not have yet.

Related Issues or PRs

#3217 (EIP-8070 section). Follows #2948, #3344, #3359.

Checklist

• Ran fast static checks to avoid CI fails, see Code Standards & Verifying Changes: just static
• PR title has the form <type>(<area>): <title>, where <type> and <area> come from an appropriate C-<type>, respectively A-<area>, label. The title should match the target squash commit message.

Cute Animal Picture

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.49%. Comparing base (d4c28bf) to head (38e234b).
⚠️ Report is 9 commits behind head on forks/amsterdam.

Additional details and impacted files

@@                 Coverage Diff                 @@
##           forks/amsterdam    #3622      +/-   ##
===================================================
+ Coverage            94.47%   94.49%   +0.02%     
===================================================
  Files                  624      624              
  Lines                36969    36964       -5     
  Branches              3334     3331       -3     
===================================================
+ Hits                 34925    34930       +5     
+ Misses                1443     1435       -8     
+ Partials               601      599       -2     

Flag	Coverage Δ
unittests	94.49% <ø> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[LouisTsai-Csie]

I'd like to understand how we verify the no-op semantics here. After sending custodyColumns = null, how do we confirm that the previously configured custody columns remain unchanged?

It seems like this state is not directly observable within our current test framework. Is this behavior verified somewhere outside this framework, and if so, where is the verification?

[LouisTsai-Csie]

The current logic allows both custody_columns to be set and custody_columns_null to be True at the same time. These two states should be mutually exclusive.

In that case, the implementation silently gives precedence to custody_columns and sends the bitmap, ignoring custody_columns_null. I don't think this combination should exist by the test framework in the first place.

For example, a test like the following should not be possible:

blobs_test(
    pre=pre,
    txs=txs,
    get_blobs_version=4,
    cell_mask=ALL_CELLS_MASK,
    custody_columns=custody_columns,
    custody_columns_null=True,
)

It would be preferable to explicitly reject this combination, e.g. with an assertion that custody_columns and custody_columns_null cannot both be set.