All the tests that follow would fail anyway, making it hard to filter
them out to avoid false positives.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54558

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54669

Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>

Whenever the Wayland output display state changes (including during
initialization), update the monitor information on the Wine side to
reflect these changes.

For now all monitors are placed at 0,0 in the monitor space, and
only the current mode is reported. These deficiencies will be addressed
in upcoming commits.

We currently support a single GPU, and a single monitor for each
adapter.
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>

Allow passing NULL as the user driver to __wine_set_user_driver(),
to set the internal null user driver. This is useful for drivers
that may need to tentatively set their own user driver during setup
and reset it on failure.
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>

Try to connect to the Wayland compositor, and fail driver initialization
if we are unable to do so.
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>

Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>

Add the initial driver stub for the Wayland driver and build it
by default.
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>

Signed-off-by: Connor McAdams <cmcadams@codeweavers.com>

Signed-off-by: Connor McAdams <cmcadams@codeweavers.com>

Signed-off-by: Connor McAdams <cmcadams@codeweavers.com>

Signed-off-by: Connor McAdams <cmcadams@codeweavers.com>

Signed-off-by: Connor McAdams <cmcadams@codeweavers.com>

Signed-off-by: Connor McAdams <cmcadams@codeweavers.com>

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54690

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54690

FVWM by default uses a focus follow mouse model so the window under the mouse cursor automatically
gets focus. Windows explorer.exe and other windows managers use click to focus model. So on FVWM,
if a test changes the cursor position, it might affects other tests that rely on a specific focus
window. Restore the cursor position after sending simulating clicks to avoid affecting other tests
unintentionally. FVWM could be configured to use a click to focus model, but right now it will make
many tests starting to succeed and other tests fail. So it seems to be too big of a change. Flaky
is added to test_SetWindowPos() because this patch makes the tests succeed on Gitlab CI but the same
tests still fails for other window managers and on TestBots.

FVWM doesn't respect PPosition hints by default and tries to tile a window according to its rules.
This might break tests that require a window at a specific position. For example, when setting the
caption bar height to anything other than 18, some user32 tests start to fail. They succeeded
previously just because the caption bar and border height on FVWM totals to 18, which is happens to
be the same value used by Wine by default.