Extend CheckOffscreenObjects to handle abnormal hitboject coordinates

[Hiviexd]

Related to #31223

[cihe13375]

Some issues:

1. slider bodies and slider tails should be allowed to go beyond 512x384. They work fine in stable and widely exist in ranked maps
2. I don't think stacks should be considered. Many maps include stacks at x=0 or y=0, and with this check all of them become unrankable.
3. y>384 should be allowed; they are not clamped in stable, and there are ranked maps actually use this.
4. I see occasional false positives for objects exactly on the boundary.

[Hiviexd]

1. slider bodies and slider tails should be allowed to go beyond 512x384. They work fine in stable and widely exist in ranked maps

Slider bodies are untouched by this PR, they're handled by the offscreen check which has been working as intended for years. Re: tails, will need examples of said ranked maps to verify.

2. I don't think stacks should be considered. Many maps include stacks at x=0 or y=0, and with this check all of them become unrankable.

Any example ranked maps I can use to test against this edge case? Agree with excluding stacks from this check.

3. y>384 should be allowed; they are not clamped in stable, and there are ranked maps actually use this.
4. I see occasional false positives for objects exactly on the boundary.

Will also need example beatmaps for both points.

[cihe13375]

1~3 are quite common, here are just some random examples:

1. https://osu.ppy.sh/beatmapsets/908558#osu/4745366 00:22:426 (2) slider tail y=-13.17
2. https://osu.ppy.sh/beatmapsets/2177056#osu/4597307 02:09:234 (8) stack at x=0
3. https://osu.ppy.sh/beatmapsets/2347298#osu/5085777 01:57:826 (3) slider head y=397

Please ignore point 4 (they are indeed caused by stacks), sorry.