Characterize root cause of intermittent test failures

In TXNCXX-29, I ended up adding a 10 sec sleep after getting the bucket, and the collection, before using them in the tests. I could see we were trying to use the connections while things were going on in the background (select bucket had not returned maybe?). So – we see NO_BUCKET coming back from the server, and a few different possible errors (including a timeout) from LCB.

Spend time to get a good set of logs, characterize exactly what is happening, and file a bug in the appropriate place. I believe this was on 6.6 and 7.0 - does it happen on 6.0.x? I suspect this is why python had similar intermittent issues, as it is recreating instances and using them in a fairly tight loop. But lets look harder, when we are not trying to get the alpha out, and understand the problem better.