Report Writer MVP

Ayo thanks again for putting this stuff together. Looking amazing, you're the 🐐 . If there's any way we would wire this shit up Tuesday, that would be awesome. I am going to "try" to demo this at the meeting.

1. Wire the picked comment section and sub section to the saved comment after confirmation.

2. The note field on "Add Comment" is not actually wired to the ingestContext field.

3. When you submit a comment, instead of immediately going to the queue page, just reset the form and add a toast that says its queuing now.

4. At the top of the queue page, we should add a completed filter to show all completed comments

5. Photo captions/llocations: When the AI runs, the photo should now have a caption of some kind saved to the media. Default add this caption to the photo captions in the edit comment view.

6. In the list of suggested comments post processing, we need to add a few things.

   1. the ability to hit a button and see the entire narrative. Right now it clamps at 2 lines. add something to view the whole thing.
   2. Start with just the top 3 and add a view more button above the AI Generated Narrative button
   3. Same with the LLM one. Need to be able to view and read the whole thing.
   4. If you pick one comment and back out and pick a different one, the narrative doesn't change.
   5. The inputs for both the title and narrative are pretty buggy if you type fast. Likely need to be uncontrolled with React Native. Types a bunch of extra letters or lags.

7. Arrows dont actually save to the media. We should basically flatten the image after image processing. Not urgent. Can do this later when we optimize media.

8. IMPORTANT: Even tho I selected info/defect, the AI matches comments from both. We need to make sure the server is filtering for this or we are passing something to filter one or the other. Work with @ben on this.

9. Would be nice to have AI also fill in the comment location. Right now it fills in the photo location which is cool but we might not use the photo location. General location based on the photo locations if possible. @ben should look at this as well.

10. Add ability to discard and delete comments in the queue.

11. When adding photo captions, the keyboard covers the inputs.