Extend gap handling alignment capabilities

[ThomasKraft]

The gaps.R script and gapless_Datetimes() function create a sequence that is anchored to the minimum and maximum of the specified Datetime column. Due to minor time inaccuracies, I'm encountering frequent issues where there is a phase shift in time data sequence that occurs part way through files, e.g., 12:00:02 to 12:00:04 to 12:00:06 to 12:00:07 to 12:00:09 to 12:00:11. This results in a plot like the one below where gaps and irregular data are consistently found even during periods where data collect is actually regular:

I'd like to propose 2 alternative ideas for modifying the behavior of gap handler to handle this type of case. The first would be a procedure that does some basic phase shift detection based on some tolerance paramter to determine how far a timestamp can be from the expected sequence. Then if a phase shift is detected, align the gapless sequence based on actual timestamps instead of forcing alignment to the min/max.

Alternatively, one could simply round timestamps to the nearest epoch interval, so in the example above you would find where the phase shift starts and then round up to 12:00:02, 12:00:04, 12:00:06, 12:00:08, 12:00:10, etc.

[JZauner]

Indeed, this is an issue I have no good answer to. Many functions in LightLogR require a regular sequence in order to work properly. While it could certainly be changed to be based on a by-timepoint-difference basis, I find this unsatisfactory as well.

The phase-shifting detection you propose sounds interesting, but would it work properly if there are many such shifts? You could test it on the included sample.data.irregular dataset, that has a basic 15-second interval, but every 10th or so measurement is 16 seconds (and every 100th or so 17 seconds...) and we could see how the proposed modification would handle that. Would that be an option?

As an aside - indeed, currently if I encounter such irregular data, I either switch to a coarser interval with aggregate_Datetime(), or I round the timestamps with cut_Datetime(Datetime.colname = Datetime). In the latter case it is important to check for duplicate timestamps as a result of rounding.