Calculate best chapter strategies for quotation denormalization during analysis

Currently, we have no visibility into how often quotation denormalization is not usable on particular chapters. The only place this information is captured is in a remark on the draft. Also, we recalculate these best chapter strategies for a book every time USFM is fetched.

If we calculated these in advance during post-processing (i.e., once we have the pretranslations), we would:

• have visibility into what chapters are denormalizable (could add a warning potentially as well)
• but have to store these strategies in the model somehow;
• increase the time it takes to run postprocessing by probably a few seconds
• but decrease the time and computational expense of fetching USFM;
• and also be able to simplify the PretranslationService logic and potentially parse the USFM only once with all update block handlers.

Start by reading PretranslationService and the post-processing path that handles pretranslations. Trace how USFM fetching currently recalculates chapter strategies and how the model could expose stored results. Done means post-processing calculates and stores the strategies, while USFM fetching can use them and visibility into non-denormalizable chapters is available.