Fix break generation not accounting for concurrent hitobjects correctly

[peppy]

Closes https://github.com/ppy/osu/issues/28622.

[frenzibyte]

This part should also be changed to get the maximum end time (i.e. Beatmap.GetLastObjectTime()) instead of the last object's end time, right?

I cannot easily imagine a test case that warrants that to be applied though (I don't understand the logic quite well).

[bdach]

I don't understand the logic quite well

The check is supposed to remove manually-adjusted breaks which are placed after all objects so that they don't hang around.

This part should also be changed to get the maximum end time (i.e. Beatmap.GetLastObjectTime()) instead of the last object's end time, right?

In theory yes, in practice it probably doesn't matter because of two circumstances:

• Objects are assumed to be sorted by start time implicitly. This is also the reason why this pull even sort of works; if objects weren't guaranteed to be sorted by start time, this approach would be breakable, and instead you'd have to do something like pull start/end times of all objects into a flat list, sort them, and do a sweeping algorithm like the "balanced brackets" algo to actually determine which time intervals have no overlapping hitobjects.
• The particular check in question here also checks if the break overlaps any objects, which means that even if the last object's end time is inaccurate, the overlapping check is going to remove the break anyway.

Due to the above I'm pretty sure it's not possible to manufacture a test case that exercises this. I'll add one for posterity at least but yeah.