Recent update greatly increased GPU memory usage

[ducha-aiki]

Matching 2 images with 20000 matches works with commit https://github.com/Parskatt/RoMa/commit/69cefb130b3ec7586da792ed897adf6e801acb1a and doesn't with currentmain.

I guess that is because of removing

    torch.backends.cuda.matmul.allow_tf32 = True # allow tf32 on matmul TODO: these probably ruin stuff, should be careful
    torch.backends.cudnn.allow_tf32 = True # allow tf32 on cudnn