Context action for images to ask for description

[julien-nc]

There could be another button next to the one to delete an image in a document. This would invoke the assistant with the task type to ask questions about images (recently implemented). The assistant input form would be pre-filled with the image in question. The result could be inserted after the image.

cc @blizzz

[marcoambrosini]

This could be done with the assistant button on the right, which would take as context the current block on the left. Actions that the assistant show could change dynamically depending on the type of block: image, video, text, ...

[julien-nc]

The core:analyze-images task type can be used for this. So it should be checked whether this task type is available or not.