I just noticed that the Gregorian definition of leap years is hard-coded in datetime-fortran:
https://github.com/wavebitscientific/datetime-fortran/blob/master/src/lib/mod_datetime.f90#L1053
I guess this would prohibit the use of different calendars (e.g. noleap) - could this cause any problems, e.g. with RYF?
Note that this also means a proleptic Gregorian calendar is applied prior to the year 1582 (see COSIMA/access-om2#117).