Better logging/tracking of upgrade process

We have little introspection into the offline upgrade process, ns_server does a good job of logging about the online process when an upgraded node comes back online and into the cluster, but for the actual offline (rpm etc) upgrade there is little information that is captured.

It would be useful therefore, for the upgrade process to generate its own log file detailing the steps it took and any problems it encountered, which we can store in the normal logs place to be collected by cbcollect_info to allow debugging of upgrade issues. For example, were the couchstore files upgraded correctly is one of the important things we should be able to tell.