I wonder how to express a time span in command lines.

First, I think that the end of a time span should be exclusive. `Sps_exposure.time_exp_start` has microsecond accuracy. If the end of a time span is inclusive, it is difficult to iterate over dates without losing unfortunate records in the gaps at midnights. I mean, if a user wants to do this:

    somecommand.py --daterange=2021-01-01T00:00:00..2021-01-01T23:59:59
    somecommand.py --daterange=2021-01-02T00:00:00..2021-01-02T23:59:59
    somecommand.py --daterange=2021-01-03T00:00:00..2021-01-03T23:59:59
    :

Then, a record at 2021-01-01T23:59:59.999999 will be lost. I don't know how well SQLite3 works, but I know some poor programs that occasionally output 2021-01-01T23:59:60.000000 by rounding 2021-01-01T23:59:59.9999996 (This is not related to leap seconds). So, it may be useless for the user to write

    somecommand.py --daterange=2021-01-01T00:00:00..2021-01-01T23:59:59.999999
    somecommand.py --daterange=2021-01-02T00:00:00..2021-01-02T23:59:59.999999
    somecommand.py --daterange=2021-01-03T00:00:00..2021-01-03T23:59:59.999999
    :

because these commands still cannot catch 2021-01-01T23:59:60.000000. The user cannot simply say 2021-01-01T23:59:60 because python's datetime.datetime does not permit 60 for the second.

Second, if I make the end of a time span exclusive, then it is confusing to keep using ... Should I use '/' as in https://en.wikipedia.org/wiki/ISO_8601 ?

What about having separate --begin and --end arguments? Both could default to unset, specifying an open range.

I added -visit-start, visit-end, date-start, and -date-end. Could you review the change?

Very nice work, thanks!

Merged to master. Thank you.