consistent time unit

[clayton33]

I've recently started to process a handful of our Sea Explorer missions and we (@dankelley and @richardsc) noticed that the unit for time varies between the timeseries, profiles, and grid netCDF files. Currently the units are the following from the associated function

function	time unit
seaexplorer.raw_to_timeseries	milliseconds since 1970-01-01T00:00:00+00:00
ncprocess..extract_timeseries_profiles	seconds since 1970-01-01T00:00:00+00:00
ncprocess.make_gridfiles	nanoseconds since 1970-01-01T00:00:00+00:00

Perhaps they should all be seconds since 1970-01-01T00:00:00+00:00 to follow recent changes made to the ncprocess.extract_timeseries_profiles ?

[jklymak]

Yes I agree. Sorry for the inconsistencies - I hadn't appreciated how important keeping these consistent was but it does screw up various processing if the units change so this should be standardized

[jklymak]

@callumrollo does it make sense to fix this? xarray makes these issues seamless, but I've had complaints from my Matlab users about the time changing units, and perhaps R-users have a similar issue. I think for ERDAPP ingestion it is also better to be float64 in seconds since 1970.

[callumrollo]

Yes it makes sense to impose consistency in the time units in SeaExplorer processing. We currently do it as part of our post processing. I'll work on a PR next week to standardise on float64 seconds since 1970

[callumrollo]

There are some other improvements I'd like to make to the SeaExplorer processing now that there is a stable version 1 release of polars, so it's a good time for an overhaul