Basically moves `PopoverContainer` to cover both the columns and the customisation panel, so that if the customisation panel is clicked on, the popover container will notice that and hide popovers like the mod preset popover.
Normally I would just block keyboard input from going past `ModCustomisationPanel`, but it's a little complicated here since I'm dealing with global action key binding presses, and I also still want actions like `GlobalAction.Back` to get past the customisation panel even if it's expanded.
The very base class is the wrong place for it because
`FreeModSelectOverlay` inherits from it, and that one has different
assumptions and rules than "user" selection. In particular, in non-user
selection, more than one rate adjust mod may be active, which breaks the
mod speed hotkey's basic assumptions.
Also includes a test case in `TestSceneShearedButton`s to ensure the buttons' shear factors don't change on different heights (I've encountered issues with that previously).
In the Mod selection area, the search bar's focus could be changed by pressing TAB.
However, when clicking outside of the search bar, the focus would be killed but two TABs were required to get the focus back on the search bar.
This happened because the action of clicking in an empty area would trigger the search bar to change its appearence, but not its internal state.
In my solution, I made the OnClick function aware of the search bar's state, so it would not only change its appearance, but also its state.
Now, after clicking in an empty area, there is only needed one TAB to select the search box again, as expected.
Closes https://github.com/ppy/osu/issues/27549.
I'm not entirely sure why the old solution failed exactly, but also
think it's unimportant because I think past me was an idiot and was
playing stupid games with the juggling of indices between two callbacks
with no ordering guarantees and expecting not to win stupid prizes.
I'm not sure this requires any follow-up reconsiderations of that entire
text flow API, but if opinions differ, I'll re-examine.
Starting from the end (which should be fine with the current number of columns, even on different/wider screen resolutions), and with a custom decay value when it reaches zero offset.
