v0.4.0 changelog

• Added Agent UI-inspired best practices, flow taxonomy, evidence contract, and performance/reliability metrics pages.
• Expanded quickstart, project classification, gate matrix, acceptance scenarios, and verdict guidance with runtime-backed surface evidence rules.
• Updated source traceability, navigation, README, and LLM entrypoints to include Agent UI, Agent Knowledge, Playwright, Vitest, pytest, and local star-project references.
• Bumped package and public plan schema version to 0.4.0.