fix(pict): spec-conformance fixes + comment classification

[righ]

Summary

Changes to the PICT model parser (TypeScript + Python):

1. Spec-conformance fixes — align constraint & sub-model parsing with Microsoft PICT's documented syntax (pict/doc/pict.md).
2. Comment classification — parse # comments into structured, per-field descriptions (TS only).
3. Docs — reflect the above in the PictModel reference and the Compatible PICT tool page.

1. Spec-conformance fixes

#	Issue	TS	Python
1	Numeric IN sets rejected — IN {512, 1024} threw Unexpected token in array (this is the value form used in PICT's own docs)	✅ fixed	✅ fixed
2	LIKE treated regex metacharacters as special — LIKE "4.8*" matched 4x8.1 because . leaked into the regex	✅ fixed	already correct (re.escape)
3	Sub-model order was mandatory — { A, B } without @ N was rejected; per spec the order is optional and defaults to the global order (/o)	✅ fixed	✅ fixed

• Numeric IN sets — the set parser now accepts NUMBER tokens (integers/floats), not just quoted strings.
• LIKE escaping (TS) — escape every regex metacharacter, then restore only * → .* and ? → . as wildcards, matching PICT semantics where all other characters are literal.
• Optional sub-model order — SubModelType.strength is optional and resolved to the global order at generation time (sub.strength ?? this.strength). A malformed order such as { A, B } @ x still produces an error.

2. Comment classification

Each maximal run of # lines becomes a CommentBlock, classified as either a field description (attached to the parameter directly below it, no blank line between) or freestanding documentation (separated by a blank line, or above a constraint / non-parameter line). classifyComments() is the single source of truth; CommentBlock and classifyComments are exported, and PictModel.comments exposes the blocks.

3. Docs

• docs/contents/reference/pict.mdx — documented numeric IN sets, optional sub-model order (syntax table + PICT-differences section), and LIKE literal semantics.
• docs/contents/tools/pict-online.mdx — same clarifications in the tool page's Constraints/Parameters tables.

Testing

• TypeScript: npm test → 18 suites, 179 tests passing (includes pict-comments.test.ts and new numeric/float-IN and LIKE tests).
• Python: pytest → 47 tests passing (numeric & float IN sets, order-less sub-models, malformed order, LIKE literal/wildcard).

Not included

• The alias-in-constraint-literal behavior ([SKU] = "Datacenter" matching an alias rather than only the first name) is a separate spec question left for a design decision.
• Unrelated working-tree changes (Buy Me A Coffee widget, favicons, vendored pict/ reference repo) are intentionally excluded.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.74%. Comparing base (29230ec) to head (bd37571).

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #51      +/-   ##
==========================================
+ Coverage   80.74%   81.74%   +1.00%     
==========================================
  Files          35       35              
  Lines        3142     3178      +36     
  Branches      416      425       +9     
==========================================
+ Hits         2537     2598      +61     
+ Misses        592      567      -25     
  Partials       13       13              

Flag	Coverage Δ
python	74.37% <100.00%> (+1.66%)	⬆️
typescript	89.11% <100.00%> (+0.21%)	⬆️

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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.