The inventory can be moved pre-interface tick by the tli subchange
listener, as well as post-our client tick event from script events
running.
This makes it difficult to correctly and generically move the interface,
since it would have to be moved pre-interface tick and also pre-frame.
Currently the code moves the interface pre-frame, which looks okay, but
since it is ticking the interface when it is a different position,
clicks on it do not work correctly.
We do not have events for pre-interface tick, so use forced position to
reliably set the position
Hovered overlay already accounts for render order, with the exception of
the vanilla WidgetOverlays which are drawn prior to any of our overlays
being drawn. Instead we order them in render-order, which works for the
dynamic-layer ones.
The minimap must be top_right so that our overlays layout around it
correctly, which currently causes it to be picked up first over dynamic
overlays due to the way overlay sorting is done.
Ctrl + backspace is an overwhelmingly common keybind for clearing the
word before the cursor. Meanwhile, clearing the current line is not as
common a keybind, so having an unset default is more sensible here for
users to decide for themselves what keybind they would prefer.
because WidgetOverlays in their default position render in the vanilla
position it can cause other overlays to occlude previous overlays when
they should not
Prevents an issue with clients being resizable with toolbar-less means in some environments (e.g. double clicking title bar on macos).
Honestly I'm not entirely sure why this is ignored if the window isn't visible, rather than being propagated to the window manager when the window is shown, but I guess Swing isn't that smart (like always).
I originally found the information on [this SO answer](https://stackoverflow.com/a/14884056) and it seemed to resolve the issue.
Closes#14235
