report failed live-migrations

[osfrickler]

currently host-action.py continues regardless of whether the live-migration was successful or failing. it would be better to stop on a failed migrations

[fzakfeld]

From my experience there will always be at least a few failed ones when migrating a large server with a close to three-digit number of VMs. Not doing many live migrations at once usually helps a bit (although the resource manager already waits for one to complete). I'd like the option of "waiting for X seconds, then retry for a max of Y times before giving up".

Also sometimes the memory changes to quickly - to be fair we mostly know this issue from older clouds where the networking is possibly a limiting factor. In this case, a 'force-complete' is necessary (but should be triggered manually via prompt, as this causes the instance to be halted for a longer time).