Address time zone issues for run duration calculations

[raffaelladevita]

Switch to using ZonedDateTime to build a date-time from the RCDB run start time and remove the dependence on the time zone that was making the FCup conversion from counts to nC dependent on the local time of the machine where the code is running.
Since the RCDB Java API doesn't provide a date but only the local time, the assumption that remains is that the run start time is within a day from the current time. However, using ZonedDateTime.minusDays method ensures the start time date is set correctly al in case of transition from or to day-light saving.

Tests on the last file from run 6419 that was recorded across a transition from EST to EDT show also the 1h time change is handled correctly in the calculation of the scaler integration time:

(printouts from two FCup scaler readouts

Run start time:	2019-03-10T00:24:49-05:00[America/New_York]
Current time:	        2019-03-10T05:15:16-04:00[America/New_York]
Integrated charge (nC):    566821.729 
Integrated time (s):       13827 
Average beam current (nA): 40.994

Run start time:	2019-03-10T00:24:49-05:00[America/New_York]
Current time:	        2019-03-10T05:15:17-04:00[America/New_York]
Integrated charge (nC):    566818.661 
Integrated time (s):       13828 
Average beam current (nA): 40.991