test: make the CP1252 overlap regression self-contained

[SHIN09-git]

The CP1252 report regression fails in a sparse checkout when neither the local research files nor the user's cache provide a kappa excerpt. It then asserts on an empty result set without exercising the encoding behavior it is intended to check.

Run the current standalone script in a temporary repository containing one UTF-8 README and use an explicit fresh offline cache path. Assert both the single result and its escaped excerpt. This makes the test independent of checkout contents and user cache state while preserving the coverage introduced by #1299. Only this test changes; no new mathematical result is claimed.

Validation on Windows / CPython 3.12.10, based on main 0b1cccf86e2f5132ed286a26ef0a681a495740ea:

• Before: 17 of 18 onboarding tests pass; this test fails because there are zero matches.
• After: all 18 pass with python -B -m unittest discover -s tests -p test_research_onboarding.py -v.
• Negative control: replacing backslashreplace with strict handling in a temporary script copy produces UnicodeEncodeError and makes the updated test fail.
• git diff --check passes. The full test suite and mathematical replay were not run for this test-only change; hosted CI is separate.

The production encoding fix and original test remain credited to #1299 and commit e76026571dbfa7299bf474b941c9abeb5bd556b3. Prepared with substantial OpenAI Codex assistance.

Scope: verifier-engineering / portable-replay, refined to this test fixture. The submission overlap refresh completed at 2026-10-14 07:23 UTC, covering 389 open PR records and 346 Discussion body/comment/reply records; the matching discussion described the already-incorporated #1299 fix. Local research-note coverage was limited by the sparse checkout, and code diffs were not comprehensively searched.

[Mister-Meeseeks]

Batch review at 3771b3809123c144ed71a5906670dc59364654b1.

Merged to main, preserving the original contributor commits and package attribution.

Reviewed scope: Isolate CP1252 onboarding regression from the live overlap cache and full repository using a temporary one-package fixture. Preserve production UTF-8 behavior and run combined tests with PR1623.

The pinned scoped command passed; the batch ledger contains its command, input hashes, environment and log. This incorporates the research/tooling contribution and does not change the selected κ or discharge the inherited multiplication-theorem assumptions.

Batch review, attribution and evidence. Thank you for the contribution.