The supervisord uses by slapos instance for instances was not checked by promises, so if an error happened while starting it the first time, partition was not reprocessed (this is the problem we observed when updating)

There's also a slapos proxy which was not checked by promises.

This adds the two missing promises.

While doing this, I noticed that the ~/bin/slapos wrapper can behave unexpectedly when user has $SLAPOS_CONFIGURATION in their environment. While that's probably fine when using this command interactively, when this command is used to run service we don't want to be influenced by environment, so the service running supervisord was also updated to always use the expected config file, to prevent potential problem when slaprunner is started with $SLAPOS_CONFIGURATION pointing to another config file.