fix: rename Claude skills, correct hooks template docs, trim CLAUDE.md

[aprprprr]

This PR fixes the Claude Code dogfooding issues found during review and testing.

Changes in this branch:

1. Renamed the collision-prone Claude Code skills:

   • code-review -> advocacy-code-review
   • testing-strategy -> advocacy-testing-strategy
     This avoids shadowing by commonly installed global skills while preserving the advocacy-specific workflows.

2. Added a real hooks-template.md configuration example and corrected it to match current Claude Code hook behavior:

   • uses .claude/settings.json
   • parses hook input from JSON on stdin
   • includes Edit|MultiEdit|Write
   • documents the current command-based hook model and lifecycle events

3. Trimmed CLAUDE.md from 70 lines to 48 lines by moving the desloppify guidance into a scoped rule file.

4. Updated Claude Code docs to stay internally consistent after adding desloppify.md:

   • 8 scoped rules
   • 6 skills
   • refreshed Claude-specific README entries

Issues addressed:

• Issue Claude Code: skill names 'code-review' and 'testing-strategy' shadow common global skills #1: skill name collisions with global Claude skills
• Issue Claude Code: hooks-template.md lacks actual settings.json format #2: hooks template lacked usable configuration
• Issue Claude Code: CLAUDE.md is 70 lines but README states 'under 60 lines' #3: CLAUDE.md exceeded the stated length guideline

[aprprprr]

Hello, would you be open to a small follow-up PR adding a basic GitHub Actions workflow for PR validation? I have GitHub Pro and can set it up; this repo currently has no automated checks, so it would help catch docs/config issues early.