Bulk select proposals/candidates exceeds API rate limit

[mzur]

If a user uses the Shift+Click feature to bulk-select proposals/candidates, they can easily exceed the API rate limit because each proposal/candidate is updated with a separate request.

To fix this, update the proposal/candidate API endpoints to accept a list of up to 100 updated proposals/candidates (similar to the annotation bulk API endpoint). In the frontend, the single change then sends a request with a single list item. A bulk-select action, however, now sends bulk requests, too (splitting the actual number of changed proposals/candidates into chunks of 100).

[mzur]

Training proposals are updated (one at a time) here, annotation candidates here. The respective API endpoints are here and here. The tests can be found here and here.

For local development without a configured GPU for MAIA jobs, a fake MAIA job can be created manually as follows:

• Switch to the dev-modules branch of biigle/core.
• Create a new volume.
• Run docker compose run --rm -itu0 worker php artisan tinker to enter the application console.

• Create a MAIA job where training proposals are selectable (replace the volume ID):

  $volume = Volume::find(15768);
  
  $j = Biigle\Modules\Maia\MaiaJob::factory()->create([
   'volume_id' => $volume->id,
   'state_id' => Biigle\Modules\Maia\MaiaJobState::trainingProposalsId(),
  ]);
  
  $p = Biigle\Modules\Maia\TrainingProposal::factory()->count(5)->create([
   'job_id' => $j->id,
   'image_id' => $volume->images->first()->id,
  ]);

  Alternatively, create a MAIA job where the annotation candidates are selectable:

  $volume = Volume::find(15768);
  
  $j = Biigle\Modules\Maia\MaiaJob::factory()->create([
   'volume_id' => $volume->id,
   'state_id' => Biigle\Modules\Maia\MaiaJobState::annotationCandidatesId(),
  ]);
  
  $p = Biigle\Modules\Maia\TrainingProposal::factory()->count(5)->create([
   'job_id' => $j->id,
   'image_id' => $volume->images->first()->id,
   'selected' => true,
  ]);
  
  $a = Biigle\Modules\Maia\AnnotationCandidate::factory()->count(5)->create([
   'job_id' => $j->id,
   'image_id' => $volume->images->first()->id,
  ]);

• You can find the MAIA job listed on the page http://localhost:8000/volumes/15768/maia (again replace the volume ID). There you can also delete the job if you no longer need it or want to create a new one.

The training proposals and annotation candidates won't have actual thumbnails with the faked data but they can be (bulk) selected, which is enough for this issue.