fix(reporting): stabilize DepMap HTML publication

[rsolerortuno]

Summary

Stabilizes the public DepMap Public 26Q1 HTML publication for v0.5.1.

• Keeps one canonical public index.
• Shows the top 20 productive targets per therapeutic intent.
• Generates all 331 target reports independently of index size.
• Removes the duplicate targets/index.html.
• Preserves the authoritative productive baseline.
• Preserves exact stored BRAF values:
  • antibody/IO: 0.269, rank 7
  • biomarker: 0.4825, rank 11
  • small molecule: 0.8449, rank 1
• Replaces raw Markdown-like DepMap output with readable HTML metrics.
• Keeps complete escaped machine-readable evidence in collapsed details sections.
• Adds repository-owned HTML validation and regression coverage.
• Corrects outdated README reporting documentation.

Validation

• Strict HTML preflight: PASS
• HTML files: 333
• Target pages: 331
• Preflight findings: 0
• Full test suite: 783 passed
• git diff --check: PASS
• Claude independent review: APPROVE
• AGY adversarial audit: COMPLETED
• Final test exit code: 0

Scientific boundaries

The reporting layer remains read-only with respect to productive target
membership, roles, scores and ranks.

DepMap cell-line dependency is not interpreted as clinical anti-PD-1 response
evidence. Absence of tumor-cell dependency does not invalidate an immune target.

Closes #76