Full pumphistory refresh after bolus/enact if no new BG yet

[scottleibrand]

We've seen intermittent reports in Gitter of what looks like it might be a race condition around incremental pumphistory refreshes during bolus wizard events, which can cause the carb portion of the bolus wizard record to be recorded but not the insulin portion. This presents a safety issue, so to mitigate it this PR performs a full pumphistory refresh at the end of any loop where a bolus is delivered or a temp basal is set, and there's time before the next BG comes in and we need to start the next loop. This should dramatically shrink the window in which a bad incremental pumphistory refresh could impact dosing decisions, without negatively impacting loop speed or reliability.

[scottleibrand]

Tested this and confirmed it forces a Full history refresh when needed:

Checking pump status (suspended/bolusing): {"status":"normal","bolusing":false,"suspended":false}
Sending ESC ESC ESC ESC to exit any open menus before SMBing and bolusing 0.5 units
Checking pump status (suspended/bolusing): {"status":"normal","bolusing":true,"suspended":false}
Settings less than 15 minutes old. Refreshing pumphistory because: bolused, Pump profile refreshed; Settings refresh
ed; Full history refreshed through 2018-08-14T20:29:40-07:00

[scottleibrand]

When it has to retry read_full_pumphistory:

Checking pump status (suspended/bolusing): {"status":"normal","bolusing":false,"suspended":false}
Sending ESC ESC ESC ESC to exit any open menus before SMBing and bolusing 0.2 units
Checking pump status (suspended/bolusing): {"status":"normal","bolusing":true,"suspended":false}
Settings less than 15 minutes old. Refreshing pumphistory because: enacted, bolused, Pump profile refreshed; Settings refreshed; Full history refresh failed.
Retry 1 of read_full_pumphistory
Full history refresh failed.
Listening: .No interfering pump comms detected from other rigs (this is a good thing!)
Continuing oref0-pump-loop at Tue Aug 14 20:52:28 PDT 2018
Retry 2 of read_full_pumphistory
Full history refreshed through 2018-08-14T20:51:57-07:00

[alimhassam]

I'm thinking of working on a more robust fix by addressing the FIXME: comment in the incremental read_pump_history function.
This looks like a good workaround/protection in the meantime.