[QTI] Publish raw_data QTI assessment items as QTI/IMSCP packages

[rtibbles]

Overview

Publish exercise nodes of authored QTI by packaging the stored item XML into a QTI/IMS content package, rather than rebuilding it from structured fields.

Complexity: High
Target branch: unstable

Context

• Today QTI is chosen only when a node has a free_response question; QTIExerciseGenerator (contentcuration/contentcuration/utils/assessment/qti/archive.py) builds item XML from structured fields and writes imsmanifest.xml + the zip.
• Native QTI items already hold their full XML in raw_data.
• Items reference media; dependencies must be collected from the XML (existing LocalSrcPath / get_file_dependencies mechanics).

The Change

• Route nodes whose items are all QTI (no perseus_question) to QTI packaging.
• Build the package from the stored item XML: validate (Ant's backend plus setting up front end files #2), collect referenced media, assemble manifest + zip.
• Surface validation failures during publish.

Acceptance Criteria

• A node with QTI items publishes a QTI/IMSCP package whose item XML matches the authored raw_data
• Referenced media is included and correctly addressed in the package
• Items are schema-validated (Ant's backend plus setting up front end files #2) before packaging; failures stop publish with a clear error
• The package is a valid IMSCP container of QTI 3.0 items
• Republish replaces stale archives correctly
• Tests cover packaging, media inclusion, and validation-failure handling

References

• contentcuration/contentcuration/utils/publish.py
• contentcuration/contentcuration/utils/assessment/qti/archive.py

AI usage

Architecture decided with the maintainer across an iterative session: blanket QTI type with the item XML in raw_data; XSD-authoritative validation across all sources; a legacy→QTI global migration with an API-layer dual-read; ricecooker upload delegating to the AssessmentItem serializer; and a Perseus custom-interaction contract confirmed against the QTI 3.0 specification. Claude mapped the existing publish/validation/ricecooker code, proposed the breakdown, and drafted each issue; the maintainer steered every decision and reviewed throughout.