Scan the news directory, rebuild the intelligence index from article metadata,
and persist the updated index to disk.
Starts from a fresh empty index on every call so that articles that have been
deleted or renamed are automatically pruned — no stale entries can survive.
Each article file is parsed to extract its date, type, language, and metadata.
The resulting ArticleIndexEntry objects are accumulated into an
IntelligenceIndex whose trend detections are refreshed on every call.
Scan the news directory, rebuild the intelligence index from article metadata, and persist the updated index to disk.
Starts from a fresh empty index on every call so that articles that have been deleted or renamed are automatically pruned — no stale entries can survive.
Each article file is parsed to extract its date, type, language, and metadata. The resulting ArticleIndexEntry objects are accumulated into an IntelligenceIndex whose trend detections are refreshed on every call.