Preserve study-specific THP-1 HLA typing

[iskandr]

Summary

• curate the three PMID 35051231 THP-1 arms with that study's homozygous class-I typing
  (HLA-A*02:01, HLA-B*15:11, HLA-C*03:03)
• preserve PMID 33392160's independently reported heterozygous DSMZ ACC-16 typing
• document that THP-1 HLA typing is study/sub-line-specific and add curation plus observation-join
  regressions
• bump Hitlist to 1.57.2

Source resolution

Nicholas 2022 S1 repeats A02:01, B15:11, and C03:03 in both class-I haplotypes and the Results
describe its THP-1 culture as homozygous. Ghosh 2020 reports A02:01/A24:02 and
B15:11/B*35:01 for a culture obtained as DSMZ ACC-16. Cellosaurus preserves both typing records
and the documented THP-1 discrepancy, so the truthful representation is per-study typing rather
than a registry-wide superset.

No upstream issue is warranted: the source records expose the divergent typings; Hitlist was
substituting the heterozygous superset into the Nicholas study.

Verification

• ./format.sh
• ./lint.sh
• ./test.sh — 1,192 passed, one expected warning
• pytest -q tests/test_build_smoke.py — 2 passed
• real PMID 35051231 B15:11 observation join contains neither A24:02 nor B*35:01 in exact or
  class-pool sample metadata
• corpus-wide sample_ploidy_audit() is empty

Closes #416

[coveralls]

Coverage Report for CI Build 33881810600

Coverage remained the same at 83.005%

Details

• Coverage remained the same as the base build.
• Patch coverage: No coverable lines changed in this PR.
• No coverage regressions found.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

No coverage regressions found.

---

Coverage Stats

Relevant Lines:	8132
Covered Lines:	6750
Line Coverage:	83.01%
Coverage Strength:	0.83 hits per line

---

💛 - Coveralls

[iskandr]

Merged and deployed.

• Merge commit: bb11d78a54e63b4f5ca681e67e368fcee6b42213
• Release gate: 1,216 tests passed with one expected warning
• PyPI: Hitlist 1.57.2
• Published artifacts: wheel (2,118,868 bytes) and sdist (2,279,997 bytes)