skipping observations after cals and errors

[nlourie]

We are skipping observations after cals and errors. The usual symptom is that right after, say, the midnight darks complete, the robot will load an observation and then immediately skip it and go to the next one. A strange consequence is that this seems to add a bunch of random commands to the queue which get executed at the end of the night barfing a bunch of errors into the slack stream (eg image acquisitions, rotator moves, etc).

Need to examine the logic happening right after cals, and the (somewhat messy and likely buggy) logic within the log observation and go to next methods.

[nlourie]

Excised a whole section of logic that was left over from the old days with the observation log maintained on Odin. This set of methods (log_observation and log_observation_and_goto_next) were injecting poor behavior. Now only methods within roboOperator.checkwhattodo are able to re-trigger another checkwhattodo. This avoids extra calls that overlap and cause the skips. This update has been running on sky without issue for 5 days. Merging in the fix into main.