Add a --timeout flag to allow retrying operations for a longer time

[kke]

Adds a --timeout=<duration> flag for most subcommands.

Usage:

k0sctl apply --timeout=120m

The timeout is not absolutely effective for all operations as the current version of rig does not support command timeouts (v2 does but migration #755 is still unfinished) and if a remote command hangs forever or past the timeout, it won't bite and abort the command, but instead keep waiting for it to finish.

If a timeout is not given, the pre-existing per step predefined (seems to always be either 30 seconds or 2 minutes) timeouts are be used and the functionality should remain the same.

If a timeout is given, each of those operations will be retried until the global timeout expires.

I'm not completely sure about naming it --timeout as it usually would imply that a long-running operation would be cut short when the time is exceeded, but in this case the use-case should more often seem to be to make it run longer than usual.

While making this I realized there was already the --retry-timeout ("Default timeout when waiting for node state changes") flag that could have been used for some operations. It should probably be deprecated with the addition of --timeout.