Station timing offsets for 102, 104 ref 105 are NaN for large periods

[tomkooij]

Station timing offsets for 102, 104 with reference 105 is mostly NaN between:
september 22nd, 2013 and march 15th, 2016.

Probably related to HiSPARC/sapphire#147 and HiSPARC/datastore#9

Offsets are NaN because of missing and/or bad data, but also NaN well before and well after such periodes. We should be able to (or try to) determine station timing offsets just before and just after "gaps in the data".

>>> Station(104).station_timing_offsets(105)
[...]
      (1379721600, 3.6, 2.73), (1379808000, nan, nan),
       (1380585600, 2.8, 2.93), (1382486400, nan, nan),
       (1458000000, 7.4, 1.32), (1458518400, 6.8, 0.99),
[...]

[tomkooij]

This is because of PySPARC issues HiSPARC/datastore#9 (station 102) and because of failing MPV fits for station 105 (wrong PMT Voltages, no 'gamma peak' : http://data.hisparc.nl/show/stations/105/2016/1/11/)