chore: sync claude config — hooks canon, session-model policy

[ArthurBernard]

Summary

• Resync .claude/hooks/ with the canonical copies in ~/.claude/hooks/ (the four repos' hooks had all silently diverged; behaviour is unchanged — the canon reads package_dir/base_branch from workflow.json).
• CLAUDE.md "Common conventions" mirror resynced: model policy is now "inherit the session model" (old "opus always" rule obsolete); /execute-leaf description updated (complexity drives the agent's effort, not its model); dated mirror marker added so /groom-docs can detect drift.
• .gitignore: ignore .claude/settings.local.json + local Claude state in-repo (was only covered by a machine-global git ignore — now self-contained like Fynance).

Why

• 2026-07-04 audit of the cross-repo Claude config: hook forks and stale policy mirrors were the two main drift sources. Canonical hooks + sync-claude-hooks + mirror markers make the drift detectable and reversible.

Test plan

• Config/docs only — no package code touched; pytest and ruff check dccd/ unaffected.

[ArthurBernard]

Follow-up commit on this branch: the model policy in the CLAUDE.md mirror evolved from "inherit the session model everywhere" to tiered execution — judgement (orchestration, review, pick/plan/finish) stays on the session model; plan-leaf execution runs at the tier derived from the leaf's complexity (low→haiku / medium→sonnet / high→session model), escalating one tier on failed tests/verification.