BUG: OB Status Socket denotes master descriptors as successfully published, when not correct.

[alecmuffett]

Situation:

• Have 4 workers, recently killed and restarted
• Launch OB
• OB complains "Empty descriptor received for xxxxxxxx.onion" for each of the workers
  • because workers recently restarted
• OB declares:

2017-03-04 13:35:18,191 [DEBUG]: Running publish_all_descriptors()
2017-03-04 13:35:18,191 [DEBUG]: Checking if any master descriptors should be published.
2017-03-04 13:35:18,192 [DEBUG]: Publishing a descriptor for service aaaa.onion.
2017-03-04 13:35:18,192 [INFO]: No descriptor received for instance bbbb.onion yet.
2017-03-04 13:35:18,192 [INFO]: No descriptor received for instance cccc.onion yet.
2017-03-04 13:35:18,192 [INFO]: No descriptor received for instance dddd.onion yet.
2017-03-04 13:35:18,192 [INFO]: No descriptor received for instance eeee.onion yet.
2017-03-04 13:35:18,192 [DEBUG]: We have 0 IPs, not using distinct descriptors.
2017-03-04 13:35:18,193 [WARNING]: Error generating descriptor: No introduction points for service aaaa.onion.
2017-03-04 13:35:18,194 [WARNING]: Error generating descriptor: No introduction points for service aaaa.onion.

...which is all very handy, but something about this process invalidates the "NOT YET UPLOADED" status that the status socket prints. Instead it prints a successful upload time for the master descriptor, whilst marking the workers as [offline]

Perhaps the status socket should print a few more things for each master:

1. the number of distinct IPs, and/or the number of workers represented in the last master descriptor upload
2. the time of the last successful non-zero-sized master descriptor upload, or a explanation if none
3. the time of the last attempt to assemble a new master descriptor, and status thereof?