feat(aidd-pm): add interview, release-notes, data-driven-decision and roadmap skills

[NathanGto]

🎯 What & why

Adds 4 skills to aidd-pm that the product (PM/PO) training needs and the plugin did not cover yet: interview prep and synthesis, release notes, data-driven decisions, and quarterly roadmaps.

🛠️ How it works

All four were authored with the aidd-context:04-skill-generate flow (scope, plan, write, validate) and follow the existing aidd-pm structure: router SKILL.md, then actions/, assets/, references/.

• 11-interview: builds an interview guide focused on past behaviour, then restructures a raw transcript question by question with exact verbatims, pains, needs and themes. Uncertain speaker attribution is flagged, never guessed.
• 12-release-notes: collects tickets and merged changes between two refs into a fact sheet, then drafts internal and customer variants. Every line traces to a ticket, PR or commit; unknowns stay TBD:.
• 13-data-driven-decision: interviews the PM first (decision, options, what would change their mind), gets a tracking plan approved, then builds a reproducible evidence dashboard and a decision memo. The PM writes the decision.
• 14-roadmap: builds or updates a quarterly roadmap from epics, briefs, tickets and decisions, with a team view and a stakeholder view (Mermaid + table). No invented dates or priorities.

Registration: plugin.json, plugin README, regenerated CATALOG.md, synced root README counts, and the pinned count in architecture-rules.test.js:232 goes from 48 to 52.

🧪 How to verify

• node scripts/check-architecture-rules.js, node scripts/check-markdown-links.js, node scripts/check-referenced-paths.js, node scripts/check-skill-argument-hints.mjs: all clean locally.
• node --test scripts/__tests__/architecture-rules.test.js: 30/30.
• Try /aidd-pm:11-interview on a real transcript, /aidd-pm:13-data-driven-decision on a pending decision.

⚠️ Heads-up

• 13-data-driven-decision ships default confidence thresholds (insufficient under 5 distinct voices or a single source; high at 20+ voices, 3+ source kinds, 8/10 rows confirmed). They are editable per tracking plan but are defaults, not sourced values: to challenge.
• New save locations: aidd_docs/product/interviews/, aidd_docs/product/decisions/, aidd_docs/product/roadmaps/, aidd_docs/releases/.
• 12-release-notes sits next to aidd-vcs release tagging: its description excludes cutting a tag or writing a changelog entry.
• Mermaid rendering was not run through a parser locally.

🤖 Generated with Claude Code

https://claude.ai/code/session_014y1bCnGdrXuSa4zSHo5eKu

✅ I certify

• I DO CERTIFY I READ EACH LINE OF THE PULL REQUEST BECAUSE I AM A SOFTWARE ENGINEER, NOT A AI PUPPY.