This does nothing in KWin though (works anyway), and it looks like other
compositors currently do not support this flag anyway. This makes it
however easier for compositors to fix incompatbilities by adding support
for the flag.

If this should happen, we'll just map it again. Yes, this is still bad,
as that means that after unlocking the screen the output may be still
black - but at least one can workaround that by forcing Compton to
redirect again (e.g. by alt-tabbing around). Better than revealing
screen content despite lock.

Explicitly XCloseDisplay before exit (makes sure our windows actually get properly destroyed, leaving black background and not the last password prompt on the display), and add a test case tool that breaks compton --background glx.

saver_xscreensaver: also parse 'mode' and 'selected' so one can better configure this using xscreensaver-demo

They're already mlock'd anyway, but this is still better e.g. against
physical attacks.

Helpers: remove bash dependency.

cppcheck: Fix unportable use of getpwuid() (PAM might call this too and overwrite our username buffer).

Possibly interesting use: xprop -id `./get-compositor`.

Doing so is harmless as we're OverrideRedirect anyway, but may also help
a compositor ignore our window.

saver_multiplex: fix another stuckness issue made visible by 47a634fc, and add an automated test to detect it (and other screen resize issues) in the future.

Don't run tests with a Xinerama setup because Xephyr does it wrong (root window instantly resizes to the first screen, and Xrandr info is incomplete).

Set WM_NAME, WM_COMMAND, WM_CLASS on xsecurelock's windows for easier debugging and compositor configuration.

- Annotate some intentionally unused variables with (void).

- oclint found a bug in saver_multiplex that causes unnecessary screen
  saver restarts when receiving an XRandR event without a contentful
  configuration change.