perf(cmdb): optimize CI cache initialization with batch queries

[lovvvve]

• Add _build_ci_dicts_batch() helper function to batch-load CI attributes by value table, reducing N*M queries to T queries
• Optimize CI relation loading by pre-fetching ci_id -> type_id mapping to eliminate N+1 query pattern
• Refactor cache building to group CIs by type and process in batches of 500 for better memory efficiency
• Add progress tracking and error reporting with detailed logging at each stage
• Skip already-cached CIs before batch processing to avoid redundant operations
• Add aidlc-docs to .gitignore
• Improves cache initialization performance for large CI datasets by reducing database round-trips