Don't insist that windows request nonzero height or width

[gwhitney]

Reverts the assertion concerning nonzero window dimensions in
gui/widgets/window.cpp; it turns out that some windows do request zero
dimension, and it doesn't hurt anything for them to.

Fixes #4960.

[Wedge009]

I'm just wondering if it's worth still keeping the lower-bound check, but assert that it's >= 0 rather than just > 0? I shudder to think when negative values might be used, but I suppose it could happen so could well be worthwhile to check for them.

[Wedge009]

Maybe throw in a comment about why 0-value dimensions are permitted as well.

[gwhitney]

Thanks for the suggestions, done on both counts.

[Wedge009]

Maybe I'm missing something but it looks the same as what I saw earlier. :S

Edit: oh, it's an unsigned comparison - so only an upper-bound check is necessary?

[Wedge009]

I just tested a local change to make assertion >= 0 and it seems to resolve the issue.

[gwhitney]

Dang, forgot git commit --amend does not make a fuss if you (cheeks red) happened to forget to git add your changes. (I will level with you, I am sort of sad git won out over mercurial etc. I find it fairly convoluted and error-prone, but maybe it's just lack of sustained use...)
Anyhow, all better now.

[Wedge009]

No worries at all, I'm not that comfortable with the git command-line myself although I am learning (I did start my PC days in MS-DOS, so it's not that I dislike command-line, I'm just not used to it).

It looks good to me and I basically tested it just now, so I'm happy with it. If no-one else objects or takes action on this sooner I'll merge this around this time tomorrow.

[Pentarctagon]

Vultraz does need to approve UI-related changes, since he's the UI maintainer now.

[Wedge009]

Right, didn't know that was the new process. All done now, it seems.

[Pentarctagon]

Yep, going forward the maintainer for a specific area needs to approve PRs related to their area, or for areas without a specific maintainer it falls back to gfgtdf (for code).