New training data generation functions (#351)

## Description

This PR adds all of the data generation functions we used to create new,
prod-representative synthetic data. It also fixes a bug in the
augmentation code (which at this point is largely deprecated and we
should consider deleting?) and changes some regex and normalization
parsing.

## Related Issues

Closes #305 

## Additional Notes

There is a LOT here, but I think it makes sense and is easiest to tackle
reviewing things in a certain order. A lot of the functions build on
each other and add knowledge cumulatively, so going bottom up will help
later orchestrator functions make sense. I would approach things in this
order:

- Small changes to existing code (augmentation, test_augmentation,
conftest, normalize, regex_patterns, and the one loinc data file I had
to change because an unrecognized character got written in that I needed
to fix for future searches)
- schemas.LoincStruct (just a class that holds some values, but a
validated one)
- data_curation.loinc_utils and its test file (highly modular, can be
grokked without specialized domain knowledge, all later things build on
it)
- data_curation.post_process and its test file (again, very modular,
pretty intuitive and small in scope what each function is doing)
- data_curation.data_emulation and its test file (this is the big daddy
where the magic happens, but it all relies on the utils and post
processing; the file has functions organized alphabetically, but my
recommendation is to start with all the functions that start with
leading underscores because those are smaller scale helpers; then, I'd
hit the functions that look like `get_XXX_variations`, which are the
functions that actually do data changing; then, I'd move to the
functions that start with `build_` because those are more about applying
direct formulas rather than varying small things; finally, the
orchestrator function `create_[blah blah]` just calls everything else in
logical sequence)

Author: bamader

data/snoinc_extracts/loinc_component_abbrv_syn_20260223.json

@@ -357355,14 +357355,14 @@
         "abbrv": [],
         "synonyms": []
     },
-    "Ebolavirus Bundibugyo+Reston+Sudan+Ta\ufffd Forest+Zaire RNA": {
+    "Ebolavirus Bundibugyo+Reston+Sudan+Taï Forest+Zaire RNA": {
         "code": "LP437071-6",
         "abbrv": [
             "BDBV+EBOV+RESTV+SUDV+TAFV RNA"
         ],
         "synonyms": []
     },
-    "Ebolavirus Bundibugyo+Reston+Sudan+Ta\ufffd Forest+Zaire": {
+    "Ebolavirus Bundibugyo+Reston+Sudan+Taï Forest+Zaire": {
         "code": "LP437072-4",
         "abbrv": [
             "BDBV+EBOV+RESTV+SUDV+TAFV"