feat: selectable commit message generation model (default qwen3.7-max)

[guoyongchang]

Overview

Adds a selectable commit-message generation model to the commit AI settings.

• New commit_ai_model field in GlobalConfig with getter/setter Tauri commands + HTTP route.
• generate_commit_message uses the configured model, falling back to a default (DEFAULT_COMMIT_AI_MODEL = qwen3.7-max) when unset.
• Settings UI gains a model selector in the commit section, mirroring the existing voice_refine_model pattern.
• Frontend wrappers getCommitAiModel / setCommitAiModel; i18n keys for the selector (en + zh).
• Server-side: zero changes — the model id is passed through to the existing cloud provider call path.

Verification

• cargo check ✓, cargo clippy -- -D warnings ✓, cargo test --lib ✓ (445 passed)
• npm run build ✓, node scripts/check-i18n.mjs ✓ (800 keys), npm run contracts ✓ (in sync), npx vitest run ✓ (74 passed)

⚠️ Manual verification needed

• The default alias qwen3.7-max must be registered upstream by the cloud provider. If that alias is not available, change DEFAULT_COMMIT_AI_MODEL to qwen-max (a known-good alias) before relying on the default.

Plan

docs/superpowers/plans/2026-06-20-commit-message-model-selectable.md