Add VolumeSnapshot timeout to Snapshot CRD and controller

[sklarsa]

This way, we can run SNAPSHOT COMPLETE if the volumesnapshot takes too long. The actual timeout is dependent on how large the database is and the underlying snapshot mechanism. This makes creating a default for this difficult. Maybe we should default to 0, or no timeout, and only enforce a timeout if specified by an end-user.