feat: file-size threshold and observation dedup for timeline gate

- Skip gate for files under 1,500 bytes — timeline (~370 tokens) costs
  more than just reading small files directly
- Deduplicate observations by memory_session_id (one per session)
- Rank by specificity: files_modified > files_read, fewer tagged files > many
- Fetch 40 candidates, dedup/score down to 15 for display
- Reduce default by-file query limit from 30 to 15

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

src/services/sqlite/observations/get.ts

@@ -122,7 +122,7 @@ export function getObservationsByFilePath(
   options?: { projects?: string[]; limit?: number }
 ): ObservationRecord[] {
   const likePattern = `%${filePath}%`;
-  const limit = options?.limit ?? 30;
+  const limit = options?.limit ?? 15;
   const params: any[] = [likePattern, likePattern];
 
   let projectClause = '';