Fix focus change deadlock in linux_wayland backend

[Adjective-Object]

Currently, if a game using the linux_wayland backend loses focus, it will deadlock with the following stack:

(gdb) bt
#0  syscall ()
    at ../sysdeps/unix/sysv/linux/x86_64/syscall.S:38
#1  0x00005583ee357846 in std::sys::pal::unix::futex::futex_wait () at std/src/sys/pal/unix/futex.rs:72
... (omitted for brevity)
#5  0x00005583ee2e0bf8 in miniquad::window::blocking_event_loop () at src/lib.rs:159
#6  0x00005583ee1e06b8 in macroquad::{impl#3}::resize_event (self=0x1, width=800, height=600)
    at src/lib.rs:506
#7  0x00005583ee2e341b in miniquad::native::linux_wayland::xdg_toplevel_handle_configure (
    data=0x7ffc4e5c5fb0, _toplevel=0x558428e38e40, width=800, height=600, _states=0x5584298b6770)
    at src/native/linux_wayland.rs:508
...
#15 0x00005583ee135d9d in miniquad::start<macroquad::{impl#5}::from_config::{closure_env#0}<miniquad::conf::Conf, cool_game::main::{async_block_env#0}>> (conf=..., f=...)
    at /workspaces/miniquad/src/lib.rs:412
...

What's happening here is that the wayland backend is holding onto the mutex guard for the native_display() when it dispatches the resize event to event listeners.

This hits the EventHandler impl for Stage in macroquad 0.4.13, which attempts to re-lock the native_display() while it is already locked by the wayland backend
https://docs.rs/crate/macroquad/0.4.13/source/src/lib.rs#506

The fix here is to guarantee that the native display is released before dispatching events.
This looks like it was already done for the payload.decorations codepath, but not for the event handlers. I can't see a motivation for continuing to hold the lock past where we read the last data off the display, so I've simply moved the manual drop up to always execute.

[not-fl3]

Thanks for PR!