add mode to preserve GPS timestamps rather than align them to the expected sample period

[garyjg]

Right now the timestamps of contiguous scans are shifted to account for drift between the ADC clock and the GPS clock, so the hot film sample times are exactly $1/f$ seconds apart. However, now I'm thinking it might be better to preserve the GPS timestamp instead. Then the data user can decide whether to account for the drift or ignore it. Likewise, rather than breaking the output files where there are interruptions in the data, leave it up to the data user to detect gaps. Early on when exporting the data it seemed to make sense to maintain the exact timestep artificially, to simplify the ingest of the csv text data, but now that seems arbitrary and unnecessary.