Use faces API info to make sure all faces are displayed

[jschwalbe]

Ok this is a bit of dreaming, and I'm sure will take some work (aka not happening soon), but hear me out. :)

Make use of the Face Bounding Boxes data

Per https://immich.app/docs/api/get-asset-info, there is asset info that tells where the faces are detected (people.faces.boundingBox<X/Y>).
This would be helpful for those of us using the image_fit: cover setting. Perhaps a new setting could be created, such as cover_faces, which would work to cover the largest amount of the screen while still keeping all detected faces in view.

I could also see this data being used for a Ken Burns style pan effect.

[jschwalbe]

pseudoCode to follow:
allFacesX1=min(people.face[x].boundingBoxX1)
Explanation: the API gives X1, X2, Y1, Y2 coordinates for each face's bounding box. So we could iterate over each face detected, and notate the smallest X1 of all of them, and that's going to be the leftmost pixel that contains no more faces to the left of it.

so,

• smallest X1 is the leftmost bounding box
• largest X2 is the rightmost bounding box
• smallest Y1 is the top bounding box
• largest Y2 is the bottom bounding box

edit: btw I hope you don't think I'm talking down to you. This is all theoretical for me as I don't really know how to code and I'm just trying to make the path as straight as possible for you!