Show official python/typing conformance results (Basilisk sole 100%) + licensing audit

[MelbourneDeveloper]

TLDR

Documentation-only PR: surface Basilisk's official python/typing conformance result — now listed upstream and the only checker at a perfect 100% — across the site and READMEs, and complete the third-party licensing/attribution audit (Astral/Ruff first).

What Was Added?

• website/src/_data/conformanceOfficial.js — a pinned, dated snapshot of the official python/typing results leaderboard (all 7 graded checkers). pct is derived from pass/total so a typo can never desync the percentage from its fraction; every value is transcribed verbatim from results.html.
• pycroscope (the 7th graded tool) added to the homepage comparison tables (EN + zh) and the /docs/comparison leaderboard — previously omitted, so the tables now show every checker's official conformance score.
• Astral / Charlie Marsh credit in the site footer (base.njk + styles.css), rendered site-wide in EN + zh.
• Acknowledgments sections in README (EN/zh), README-pypi.md, and the VS Code extension README (Astral/Ruff first).
• PEP 639 license metadata (license = "MIT", license-files) in pyproject.toml so the Ruff copyright travels with the PyPI wheel.

What Was Changed or Deleted?

• NOTICES / THIRD-PARTY-LICENSES / vscode-extension/NOTICES — corrected the Ruff copyright to Copyright (c) 2022 Charles Marsh, pinned rev 7c645a9 (== release 0.15.17), documented the full 15-crate ruff_* family, and added the permissive-tree summary (only file-level copyleft: inferno CDDL-1.0, colored MPL-2.0, both linked unmodified; no GPL/AGPL/LGPL-only).
• READMEs — restore the "the only Python type checker with a perfect 100% score" claim now that it is substantiated by the official results.html, and link the claim to that source (Documentation Honesty: a comparative claim carries its authoritative source).
• Conformance references re-stamped c94dfce → f051625 (still 141/141, 0 false positives) in conformance_report.json, README, and CHECKER-ARCHITECTURE-SPEC.md to match the upstream tip the gate now scores against.

How Do The Automated Tests Prove It Works?

• Website build + drift guards: npm run build clean (211 pages); gen_rules_reference.py --data diff clean (rules.json in sync); gen_conformance_reference.py --check passes (README/spec stamps in sync with conformance_report.json).
• Website e2e: Playwright navigation + CLI-screenshot smoke suite — 24 passed / 4 skipped (desktop + mobile). The rendered pages show all 7 official scores (100 / 99.6 / 97.9 / 96.8 / 92.2 / 82.3 / 77.3 %).
• Conformance gate: the unmodified official python/typing scorer at f051625 reports 141/141 = 100.0%, 0 false positives (score.py --gate PASS).
• Figures verified against source: every conformance number checked against the live results.html @ f051625 (Basilisk is the sole tool at 100%); pycroscope's facts (Python, Apache-2.0, no LSP/debugger/editor extension) verified against its repository.
• No code changed (docs/metadata only): lint (clippy + eslint + deslop), Zed, VS Code VSIX (499 tests, 93% cov), Neovim (0 failures, 44% cov), Shipwright binary gate, and the release build all pass locally.

Spec / Doc Changes

docs/specs/CHECKER-ARCHITECTURE-SPEC.md conformance stamp updated to f051625 (141/141, 100%, 0 FP) — kept in sync by the gen_conformance_reference.py --check drift guard.

Breaking Changes

• None