fix: skip empty MHC items and add similarity threshold (closes #7, #8)

[majiayu000]

Description

This PR addresses two related issues with MHC (Mental Health Catalogue) matching:

1. Issue Remove empty items from MHC #7 - Remove empty items from MHC: Empty or whitespace-only MHC questions are now skipped during matching. This prevents invalid items from being matched to user questions.

2. Issue Don't match to MHC items if similarity is too low #8 - Don't match to MHC items if similarity is too low: Added a new mhc_min_similarity parameter that allows filtering out matches below a similarity threshold. Default is 0.0 for backward compatibility, but can be set higher (e.g., 0.5) to filter unrelated matches.

Changes

• Modified matcher.py to build a mask of valid MHC questions and skip empty/whitespace items
• Added mhc_min_similarity parameter to both match_instruments_with_function and match_instruments
• Added comprehensive unit tests for the new functionality

Testing

• All 108 existing tests pass
• Added 5 new tests specifically for MHC filtering:
  • test_empty_mhc_questions_are_skipped
  • test_whitespace_only_mhc_questions_are_skipped
  • test_low_similarity_no_match
  • test_high_similarity_match
  • test_threshold_filters_unrelated

Usage

# Default behavior (backward compatible)
match_response = match_instruments([instrument], mhc_questions=mhc_questions, ...)

# With similarity threshold to filter unrelated matches
match_response = match_instruments([instrument], mhc_questions=mhc_questions, 
                                   mhc_min_similarity=0.5, ...)

Fixes #7, #8

[woodthom2]

Thanks @majiayu000 , please give me a couple of weeks to check this over

[woodthom2]

Thanks so much @majiayu000 !