add per column summary statistics to the inventory resource

[lindsaywiard]

Adds summary to each Column, populated at completion time.

• New summarize.py builds the stats graph
• save_parquet / save_parquet_replace replaced by *_with_summary variants that return stats alongside the path
• Stats write back to Firestore on completion via update_status
• Single-pass regression test included

Changes

• API
  • schema.py — added ContinuousColumnSummary, CategoricalColumnSummary, ColumnSummary union, summary field on Column
• API Tests
  • test_schema.py — round-trip tests for both summary discriminator variants
  • test_duplicate_router.py, test_modifications_router.py, test_treatments_router.py — updated columns assertions to expect summary: None
• Standgen
  • summarize.py — new, _build_column_stats_graph returns lazy dask scalars per column
  • storage.py — _compute_write_and_stats fuses stats with the write; replaces save_parquet / save_parquet_replace with *_with_summary variants
  • handlers/pim.py, handlers/chm.py, handlers/modifications.py, handlers/treatments.py — swapped call sites, return columns in result dict
  • main.py — update_status gains a columns param, writes summaries to Firestore on completion
• Standgen Tests
  • test_summarize.py — new, unit tests for stat correctness, null handling, and categorical unique count
  • integration/test_storage.py — new, single-pass regression test for both write functions
  • handlers/conftest.py — new, shared column constants for handler tests
  • Handler and integration tests updated for new call signatures and summary assertions
• Lib Tests Shared Data
  • chm_lmf.json, chm_vwf.json, pim_treemap.json — added columns field

[lindsaywiard]

I can work on these merge conflicts tomorrow and update you when I get them figured out :)

[lindsaywiard]

Should be ready for review!