Python strftime time scale

[distbug]

Hi,

using CPython 3.11.3 and the latest version of hifitime (3.8.2), strftime does not work with epochs using the GPST/BDT/GST time scale.

from hifitime import Epoch, TimeScale

e: Epoch = Epoch.init_from_gregorian(year=2022, month=3, day=1, hour=1, minute=1, second=59, nanos=1, time_scale=TimeScale.GPST)
print(e.strftime("%Y %m %d %H %M %S %f %T"))

Expected:
2022 03 01 01 01 59 000000001 GPST
Actual output:
1942 02 25 01 01 40 000000001 GPST

It works if the time scale is given as UTC or TAI:

e: Epoch = Epoch.init_from_gregorian(year=2022, month=3, day=1, hour=1, minute=1, second=59, nanos=1, time_scale=TimeScale.UTC)
print(e.strftime("%Y %m %d %H %M %S %f %T"))

Output:
2022 03 01 01 01 59 000000001 UTC

Similarily to_gregorian_utc_str does not work, but that is as far as I can tell related to #209

e: Epoch = Epoch.init_from_gregorian(year=2022, month=3, day=1, hour=1, minute=1, second=59, nanos=1, time_scale=TimeScale.GPST)
print(e.to_gregorian_utc_str())

Expected:
2022-03-01T01:01:41.000000001 UTC
Actual output:
2022-03-01T01:01:22.000000001 UTC

The same wrong offset of 37 seconds is used here as mentioned in the other issue.

[ChristopherRabotin]

@gwbres Have you encountered this issue as well? I can start to look into it in two weeks, I'm unavailable before then.

[ChristopherRabotin]

@distbug , my apologies for the delay on tackling this issue. I can confirm that this has been fixed in the upcoming version 4.0.0. I've rewritten the algorithm that computes the Gregorian days from the duration in a given time scale.