BM25 prefix matching scans entire inverted index linearly

[rohitg00]

Problem

For each query term, the search iterates ALL entries in the inverted index to find prefix matches. With thousands of unique terms, this is O(queryTerms * vocabularySize).

Location

src/state/search-index.ts:78-97

Suggested Fix

Use a trie or sorted array for prefix lookups instead of linear scan.

[rohitg00]

Resolved in v0.6.0 (PR #76). BM25 prefix matching now uses binary search via lowerBound() + lazily-cached sorted term array in search-index.ts. O(log n + k) instead of O(n) full scan.