feat(contrib): add PR review and module scaffold skills

[jeanscherf]

Description

Adds two Claude Code skills to .claude/skills/ for SDK contributors. Skills are invocable via Claude Code CLI (or any compatible agentic tool: Cline, Copilot, etc.) from the repo root.

/review-pr [PR#]

Reviews a PR against 23 criteria across 6 sections:

• A: Process & Compliance — PR template, conventional commits, issue link, AI disclosure
• B: Security — credentials, internal URLs, sensitive data in code and PR body
• C: Code Quality — CI checks, version bump, type hints, naming, imports, dependencies, proto freshness
• D: API & Design — future-proofing, public API hygiene, breaking changes, pagination, telemetry
• E: Tests & Documentation — test coverage, docstring quality, module structure

Outputs a structured table report with ✅/⚠️/❌/➖ per criterion, blocking issues summary, and optional posting via gh pr review.

/scaffold-module

Generates the full standard module layout for a new BTP service integration:

• 10 files: __init__.py, client.py, config.py, exceptions.py, _models.py, py.typed, user-guide.md, plus test stubs
• Telemetry wiring: inserts Module.<NAME> into core/telemetry/module.py in alphabetical order
• Self-review phase: runs uv run ruff check on generated files and fixes issues before reporting
• Prints a checklist of remaining manual steps

Type of Change

• New feature (non-breaking change which adds functionality)

How to Test

1. From the repo root, run /review-pr 128 — verify a structured 23-criterion report is produced
2. Run /scaffold-module and provide test inputs (e.g. module=test_svc, service=Test Service, short=TS) — verify 10 files are generated, ruff passes, test passes
3. Confirm generated files are deleted after testing (the skill generates real files in your checkout)

Checklist

• My changes follow the contribution guidelines
• I have performed a self-review of my own changes
• The skills were manually tested against a real PR (PR fix(extensibility): use ConsumptionLevel for UMS destination lookup #128) and a test scaffold run
• Generated scaffold stubs pass uv run ruff check and uv run pytest
• No sensitive data (credentials, internal URLs, tenant IDs) is included
• Skills are model-agnostic via tools and compatibility frontmatter fields