Avoid using duration of 0 for non-animated cursors

[bbb651]

This fixes a crash with wayland-cursor and anvil, see: Smithay/smithay#1653
I'm not sure which behavior is "correct", the exact behavior seems to be undocumented but this reduces possible crashes.

(I edited action/build/dist/main.js directly because it's a single character and invoking pnpm updated the lockfile version and generated a bunch of noise)

[phisch]

Would this even be necessary if this is also fixed in smithay?

It's been a while since I read up on this, but I believed that other themes set 0 in case if there is only one single frame as well.

Edit: also, I have been using this with niri, which is smithay based, for about a year now. Why does it only affect anvil?

Edit 2: lets discuss this over on the smithay repo. Nobody except me uses this cursor theme builder anyway 😛

[bbb651]

Edit: also, I have been using this with niri, which is smithay based, for about a year now. Why does it only affect anvil?

Niri has a check for it.

Would this even be necessary if this is also fixed in smithay?

Maybe? Since it's ambiguous I imagine there are more cases, it doesn't help seemingly popular cursor themes don't do that e.g. Adwaita (based on it not triggering the crash it's annoying to check since it's a binary format and their build script is weird, I hate Xcursor). On the other hand 0 feels nicer and better communicates the intent of "not animated", and it feels like it should be a tiny bit faster in e.g. niri where there's an early exit but I imagine it's within margin of error or very negligible, would be interesting to bench.

[phisch]

I also dislike xcursor and existing generators. That's why I wrote an encoder for this myself: https://github.com/phisch/cursor-theme-builder/blob/master/src/lib/x11/xcursor.ts

The field used is called a delay, which is why I'm using 0 here. The field denotes the delay from the last sprite to this sprite. 0 is the only value that makes sense here imo.