feat: track per-user tool access times with in-memory cache and periodic DB flush

[aptend]

What type of PR is this?

• feat (new feature)
• fix (bug fix)
• docs (documentation)
• style (formatting, no code change)
• refactor (code change that neither fixes a bug nor adds a feature)
• perf (performance improvement)
• test (adding or updating tests)
• chore (maintenance, tooling)
• build / ci (build or CI changes)

Which issue(s) this PR fixes

N/A — new feature

What this PR does / why we need it

Adds per-user tool usage tracking: an in-memory map records the last access time for each (user_id, tool) pair, flushed to DB every 10 minutes. A query API returns results directly from memory with zero DB overhead.

Changes

memoria-api/src/auth.rs

• New ToolUsageBatcher struct: in-memory HashMap<(user_id, tool), (DateTime, dirty)>
  • mark_used() — records access with current timestamp
  • get_user_tool_usage(user_id) — returns all tool timestamps from memory
  • rebuild_from_db() — restores cache from mem_tool_usage table on startup
  • flush() — batch upserts dirty entries to DB; only clears dirty flags after all writes succeed (retry-safe)
• spawn_tool_usage_flusher() — background task, 10-minute interval
• AuthUser extractor reads X-Tool-Name header, filters empty values, calls mark_used on both auth paths

memoria-api/src/state.rs

• AppState gains tool_usage_batcher: Arc<ToolUsageBatcher>
• init_auth_pool() calls rebuild_from_db() then starts the flusher

memoria-api/src/lib.rs + routes/memory.rs

• GET /v1/tool-usage — returns current user tool access times from memory

memoria-storage/src/store.rs

• migrate() creates mem_tool_usage table with PRIMARY KEY (user_id, tool_name)

Testing

• 3 unit tests (auth::tests): mark/query, overwrite updates time, empty tool filtered
• 1 e2e test (test_tool_usage_tracking): empty state, recording, empty header ignored, multiple tools, user isolation
• INSERT ON DUPLICATE KEY UPDATE verified against MatrixOne directly