Retry of all OpenStack client calls

Sahara uses a bunch of OpenStack clients to communicate with other OpenStack
services. Sometimes during this clients calls can be occurred unexpected
errors that lead to Sahara errors as well. If you make a lot of calls, it may
not be surprising if one of them doesn't respond as it should - especially for
a service under heavy load.

You make a valid call and it returns a 4xx or 5xx error. You make the same
call again a moment later, and it succeeds. To prevent such kind of troubles,
all clients calls should be retried. But retries should be done only for
certain error codes, because not all of the errors can be avoided this way.