New endpoint (or filters) to delete ML results per worker version

A new endpoint (or filters on an existing endpoint) should be available to delete:

• all transcriptions produced by a worker version
• all transcriptions+entities produced by a worker version
• all ml results produced by a worker version

The filters would be:

• a unique worker version
• a combination of ml results type (transcription, entity, classification)

An async job would do the deletion.

See https://gitlab.com/arkindex/requests/-/issues/174

A frontend MR would of course be needed to expose that endpoint

⤵ V1 would only be to delete EVERYTHING produced by a worker version, under a specific path. So the backend would only take 2 parameters:

• corpus_id (required)
• worker_version_id (required)
• parent_element_id (nullable)

Concerned ML results to delete are:

• elements
• transcriptions
• classifications
• entities (bonus)

⚠ Only a corpus admin can do that