IFC-751 save diff in smaller pieces to prevent OOM

[ajtmccarty]

fixes #4511
IFC-751

refactors logic to save an enriched diff so that it uses multiple separate queries instead of one big one to prevent an out-of-memory error when loading a gigantic serialized diff into the database driver

new process is

• create new enriched diff root nodes (one for main -> branch, one for main -> main)
• load diff nodes in batches of 100 (one query per batch)
• link hierarchical nodes with the diff

[ajtmccarty]

the changes in here are mostly just refactoring what used to be in save_query.py, but I also renamed the file, so git counts this as all new stuff, which it is not

[dgarros]

Please can you add a fragment for the changelog before merging the PR

[codspeed-hq]

CodSpeed Performance Report

Merging #4544 will degrade performances by 92.09%

_{Comparing ajtm-10042024-save-big-diff-IFC-751 (4cb5d48) with stable (01e37a6)}

Summary

❌ 3 regressions
✅ 7 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	stable	ajtm-10042024-save-big-diff-IFC-751	Change
❌	test_get_schema	197.7 ms	805.1 ms	-75.45%
❌	test_base_schema_duplicate_CoreProposedChange	1.5 ms	18.4 ms	-92.09%
❌	test_schemabranch_process	347 ms	2,925.1 ms	-88.14%