Quit instead of shutdown on signals

[yol]

TMSG_QUIT was indavertently changed to TMSG_SHUTDOWN when the signal
handling was modified. Change back to TMSG_QUIT - the application should
not cause the machine to power down upon receipt of SIGINT or SIGTERM
after all.

Slipped through in #15730 unfortunately

[MilhouseVH]

I ran some tests in LibreELEC (without our SIGTERM patch) and the changes in this PR seem fine, TMSG_QUIT is much more polite than TMSG_SHUTDOWN! :)

I did briefly test if we still need our SIGTERM and I occasionally (but not always) saw this behaviour when testing this PR (without the LibreELEC SIGTERM patch):

PR15730+PR15776, without LibreELEC SIGTERM patch, systemctl stop kodi:

Mar 19 19:03:43 NUC systemd[1]: Stopping Kodi Media Center...
Mar 19 19:04:13 NUC systemd[1]: kodi.service: State 'stop-sigterm' timed out. Killing.
Mar 19 19:04:13 NUC systemd[1]: kodi.service: Killing process 1117 (kodi.sh) with signal SIGKILL.
Mar 19 19:04:13 NUC systemd[1]: kodi.service: Killing process 1121 (kodi.bin) with signal SIGKILL.
Mar 19 19:04:13 NUC systemd[1]: kodi.service: Main process exited, code=killed, status=9/KILL
Mar 19 19:04:13 NUC systemd[1]: kodi.service: Failed with result 'timeout'.
Mar 19 19:04:13 NUC systemd[1]: Stopped Kodi Media Center.

In this case kodi.bin would have been killed, without saving guisettings.xml and possibly leaving any CEC-connected devices in an unknown state.

I think the purpose of our SIGTERM patch is to force the update of guisettings.xml and ensure (as much as possible) a clean CEC shutdown whenever kodi.bin is instructed to stop. I'll run more tests tomorrow with/without our SIGTERM patch, but this PR looks good to go.

[Memphiz]

Really odd. When I tested that other PR it definitly did not shutdown my computer...

[yol]

@Memphiz My guess is that it only really shuts down in standalone mode (it didn't shut down for me either)