[sandragjacinto]

Tutor file routes (/tutor/files/content, /tutor/mas_files/content):

• at most 5 files (400 TOO_MANY_FILES) of 10 MB each (413 FILE_TOO_LARGE), checked before extraction from the size Starlette records, or by reading in 1 MB chunks when it is unknown; the client allows 3 files of 5 MB
• extracted text capped at 150,000 characters shared between the files, with a warning log when it is cut
• lang is Literal["en","fr"] on /files/content, /syllabus and the feedback body; /mas_files/content also accepts "english"/"french", which the MAS client sends

Search:

• nb_results le=100 on /search/collections/{collection}, /tutor/search_extracts and EnhancedSearchQuery
• relevance_factor in [0, 1], influence_factor in [0, 10]
• /search/documents/by_ids takes list[UUID] (a non-UUID was a DB 500) and, like the bibliography export, at most 500 ids
• search_multi_inputs takes at most 10 queries (400 TOO_MANY_QUERIES, summaries max_length=10) and runs at most 4 at a time behind a semaphore

openapi.json regenerated: inputs tightened only, plus documented 400/413.

Description

Why?

How?

Screenshots (if appropriate):

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)

Checklist:

• My code follows the code style of this project.
• My code is tested.
• I have updated the documentation accordingly.