Minor iconforge fixes

[tigercat2000]

Just fixes a width() call where there should be height()

[ZeWaka]

@itsmeow please advise

[itsmeow]

Just need to run a tracy profile on this and compare the performance and it's good with me

[tigercat2000]

Added another fix for another bug I found: for y in 0..std::cmp::min(image.width(), other_image.width()) only works for square icons. Using height works for both square and non-square icons.

[itsmeow]

So turns out this is way horrible for performance and is likely related to having massive spritesheet file sizes and no page cache on the Linux runners? I'm not sure if merging this is a good idea as they work fine for spritesheets that aren't massive and this adds seconds of overhead to small spritesheets that don't have issues. This needs to check for a certain size (what that size is, I have no idea)

[ZeWaka]

Just checking in, is all this code still needed?

[itsmeow]

For some use cases yes. It adds a little bit of overhead, but it's acceptable. The bug fix is necessary.