Marker to indicate bootstrap has failed is not always created

[Simon Fels]

We have seen cases where the bootstrap process has failed like

Mar 06 02:27:36 ubuntu anbox-cloud-appliance.bootstrapper[6983]: Error: Failed to run bootstrap process, you can find more information at /var/snap/anbox-cloud-appliance/com>
Mar 06 02:27:36 ubuntu anbox-cloud-appliance.bootstrapper[6983]: Error: exit status 1
Mar 06 02:27:36 ubuntu systemd[1]: snap.anbox-cloud-appliance.bootstrapper.service: Main process exited, code=exited, status=1/FAILURE
Mar 06 02:27:36 ubuntu systemd[1]: snap.anbox-cloud-appliance.bootstrapper.service: Failed with result 'exit-code'.

but the marker in /var/snap/anbox-cloud-appliance/common/.bootstrap_failed was not created causing a subsequent `anbox-cloud-appliance destroy` to fail as it misses both the .bootstrap_done and .bootstrap_failed marker. This leads to the following situation

ubuntu@ubuntu:~$ sudo anbox-cloud-appliance destroy
Error: appliance not deployed
ubuntu@ubuntu:~$ sudo anbox-cloud-appliance init
Error: appliance is already initialized

We need to ensure the .boostrap_failed marker is always created, independent of what the bootstrap.sh script does.