feat(search+docs): trigram fuzzy search, brand/country filters, developer docs

Search:
- migration 0009: trigram GIN index on brand.name + btree on country_of_origin
- SearchProducts: typo-tolerant word_similarity matching (>=0.42) on top of
  ILIKE substring + barcode; new brand/country filters; rank by
  similarity * (0.5 + quality_score). Response gains country_of_origin,
  quality_score and per-result relevance score.
- public search UI: brand/country filter inputs; show country in results

Docs:
- serve embedded OpenAPI 3 spec at GET /api/v1/openapi.json (not rate limited)
- ApiDocs page: auth + rate-limit section, updated search params/response
- docs/api.md developer guide

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

migrations/0009_search.down.sql

@@ -0,0 +1,2 @@
+DROP INDEX IF EXISTS idx_product_country;
+DROP INDEX IF EXISTS idx_brand_name_trgm;

migrations/0009_search.up.sql

@@ -0,0 +1,9 @@
+-- Search upgrade: trigram-based fuzzy matching + brand/country filters.
+-- pg_trgm and the product.name GIN index already exist (see 0001). Add a
+-- matching trigram index on brand.name so brand filtering / fuzzy brand
+-- lookups can use an index instead of a sequential scan.
+CREATE INDEX IF NOT EXISTS idx_brand_name_trgm ON brand USING gin (name gin_trgm_ops);
+
+-- country_of_origin is filtered by exact/prefix match; a plain btree index
+-- keeps that cheap as the catalog grows.
+CREATE INDEX IF NOT EXISTS idx_product_country ON product (country_of_origin);