I labelled this issue Writing Flow, because of the following:

Imagine I'm writing, hit enter, get a new block, keep writing, all looks good. Decide to insert an image and add a caption. I expect that adding a caption and hitting ENTER gets me a new Paragraph block to keep writing. Instead, it dismisses the keyboard and no new block is added, so I need to figure out what happened for a second and then if I want to keep writing I need to add a block myself.

I think we could try making it add a new default block, to keep the writing flow going.

On the Web, ENTER does not have a special treatment but only adds a new line to the caption:

wdyt?