Refactor jobs pipeline to hydrate data from snapshots

### Current
Data hydration works as follows:
1. Fetch data from source inputs (PDFs, cheminements.txt, APIs)
2. Parse the data
3. Save it to database

### Goal
Change this pipeline so hydration happens from snapshots instead of directly from source inputs:
1. Save data as a snapshot
2. hydrate the backend from that snapshot only

This should make the data aggregation pipeline easier to maintain.

(Attributes `planificationPdfJson` and `horaireCoursPdfJson` was a good start but never fully implemented, remove it and create table.s containing snapshots)