enforce_same_original_doc_id parameter & improved text pair metadata

[ArneBinder]

This changes construct_text_pair_coref_documents_from_partitions_via_relations() to add the following to metadata:

• original_doc_id: doc id of the original document where the text is from
• original_doc_id_pair: doc id of the original document where the text_pair is from
• original_doc_span: a dict with start and end keys indicating where the text is from with respect to the original doc
• original_doc_span_pair: a dict with start and end keys indicating where the text_pair is from with respect to the original doc

Also, this gets correctly propagated inadd_negative_coref_relations(), when available.

Finally, this adds the parameter enforce_same_original_doc_id to add_negative_coref_relations(). If enabled, only negatives where both text and text_pair are from the same original document, get created.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 84.61538% with 2 lines in your changes missing coverage. Please review.

Project coverage is 95.73%. Comparing base (a307df8) to head (2997307).
Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
src/pie_modules/document/processing/text_pair.py	84.61%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #121      +/-   ##
==========================================
- Coverage   95.76%   95.73%   -0.03%     
==========================================
  Files          61       61              
  Lines        5048     5061      +13     
==========================================
+ Hits         4834     4845      +11     
- Misses        214      216       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.