Due to changes in how we'll run storage migration, we need to change the handling of contents of version file.

This will result in making a more logical separation between backup/restore and version managements procedures.

$DATA/version will be used to block the start up in case of version skew problems. Only exception is taking a backup of 4.13 MicroShift because it's triggered by missing version file.

Backup/restore part will happen at the very start of microshift run, just as it is right now, it end up simplified due to this change. General logic might look like:

• backup if "healthy" in health.json
• restore if "unhealthy" in health.json

• restore if system rolled back

This is a PR that introduces new way of running migration: https://github.com/openshift/microshift/pull/2019