avoid setting duplicate cursor image

Reference: https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3436
Based on: https://gitlab.freedesktop.org/wlroots/wlroots/-/merge_requests/3595

destroy renderer and allocator in cleanup()

fix bad condition

fix various segfaults when selmon is NULL

change cursor surface in internal calls of motionnotify()

and call motionnotify() after unmapping a client
and when focusing a client

Fixes: https://github.com/djpohly/dwl/issues/308

remove unused variables

rework Makefile

- allow user to use a different pkg-config binary
- restore almost all (and add other ones) warning flags (-Werror is not set)
- $(XWAYLAND) is added to our CPPFLAGS
- remove useless comments
- don't generate idle-protocol.h (not used)

make more permissive exclusive focus

now you can call focusstack() while a layer surface is focused and when it gets
unmapped the newly focused clients will be actually focused

now really fix clients not being focused after unmapping a layer surface

fix clients not being focused after destroy a layer surface

Revert "remove useless check of `m` in arrangelayers()"

This partially reverts commit d14ee99661e420c8e3d8fcc6791429a0e08bd915.

Checking if the output is enabled is necessary to avoid a segfault later

change the default term to `foot`

it is closer to the suckless philosophy (foot implements things that the suckless
guys would say that should be done by tmux or something else, but I have no
desire to create a new terminal emulator, and the best fit is foot)

also alacritty uses +100Mb of memory, more that dwl itself (~90Mb)
and foot only ~20Mb

remove useless check of `m` in arrangelayers()

it is supossed to avoid a crash when we pass m=NULL as argument
however it would crash anyway and also because we should not pass m=NULL

fix use of loop initial declaration

forbidden by the suckless style

fix unmanaged clients not being unlinked from the commit listener

fix segfault when unlocking swaylock on two monitor setup

wlr_*_surface_from_wlr_surface() can return NULL if the surface
is being destroyed

Fixes: https://github.com/djpohly/dwl/issues/305

don't try to move outputs when its x,y hasn't change

when using wlr_output_layout_move() wlroots internally
change the state of the output to manually configured and
when updating the layout these outputs aren't ignored by
wlroots, leaving us at our own

prior run the startup command start the backend

this allow use clients like wlr-randr in the startup command

make sure the parent is mapped prior set monitor and tags

use wl_container_of() instead of data in some functions