Select the white area using color selection, invert selection, then crop to selection. If not enough of the border is selected, increase the selection threshold.
You don't have to worry about white areas within the image, such as the fountain, because the crop is rectangular. However images with white across an entire edge, such as blown out skies, may ...
If you only have the .jpg files, likely you cannot pull very much from the image even with special recovery software. If you took the .NEF/.CR2 'raw' images, I could definitely give you a hand pulling the picture out. Unfortunately, JPEG compresses the image for storing and removes and raw data - sort of like a picture of your picture.. Sorry to hear that!
Possible solution, using the ofn-extract-objects script found here(*):
Make a multi-rectangle selection, one rectangle around each object (shift-click to add rectangles)
Copy/paste the selection to a new layer: Ctrl-C, Ctrl-V, Ctrl-Shift-N. This layer will be mostly transparent, except where you selected things.
Layer>Extract objects>Extract object to ...
Yes. Give each layer a layer mask. Where the layer mask is white, the layer shows. Where the layer mask is black, it is invisible. Gray values between black and white make the layer blend with that opacity.
So make each layer mask the appropriate triangle of black vs. white and you'll be set.
I'll answer your four questions in a different way that I think makes more sense, which will hopefully also answer each one.
Adobe RGB isn't "better" than sRGB. More specifically, it is better in some respects, but not others.
Its coverage of the human-visible chromaticity diagram is slightly better than sRGB. But many programs don't understand colorspace, ...