feat(analysis): add subgroup similarity maps and policy count visualizations

Optimize visualizations with improved year window and color scales

Fix year filtering and improve color scales across visualizations

Regenerate analysis outputs with full strategy similarities

Finalize full dataset analysis with robust date handling

docs: add comprehensive setup instructions for R and Python dependencies

Remove duplicate data file: switch from DTA to CSV-only

Update documentation: abstract embedding with chunking

Improve abstract embedder: add chunking for translation and embedding

Refactor to abstract-based embedding pipeline

Move remaining log files from root to data/temp/

Update file organization: move intermediate files to data/temp, clarify conventions in CLAUDE.md

Update analysis to use all 1,327 policies with embeddings

Remove main_embeddings.py; make main.py the sole entry point for full pipeline

Change default --limit to None (all policies) in generate_embeddings.py