Provide support to extract ESMF_Time and ESMF_TimeInterval from an Hconfig

[tclune]

This functionality is already implemented in MAPL and we will do a pull request to get this added to the ESMF code base.

The only possibly difficult aspect is just which file these should be implemented in. (@oehmke ) I've not looked yet, so maybe it will be obvious, but best if ESMF provides guidance.

The other question I have at the moment is whether you will require unit tests. The implementations are almost trivial wrappers on existing functionality, so I'm hoping we can skip that bit.

[tclune]

For reference, the MAPL implementations are as follows:

   function as_time(hconfig, unusable, keystring, index, rc) result(time)
      type(esmf_Time) :: time
      type(esmf_HConfig), intent(in) :: hconfig
      class(KeywordEnforcer), optional, intent(in) :: unusable
      character(len=*), intent(in), optional :: keystring
      integer, optional, intent(in) :: index
      integer, optional, intent(out) :: rc

      integer :: status
      character(:), allocatable :: str

      str = esmf_HConfigAsString(hconfig, keystring=keystring, index=index, _RC)
      call esmf_TimeSet(time, str, _RC)

      _RETURN(_SUCCESS)
   end function as_time

   function as_timeinterval(hconfig, unusable, keystring, index, rc) result(interval)
      type(esmf_TimeInterval) :: interval
      type(esmf_HConfig), intent(in) :: hconfig
      class(KeywordEnforcer), optional, intent(in) :: unusable
      character(len=*), intent(in), optional :: keystring
      integer, optional, intent(in) :: index
      integer, optional, intent(out) :: rc

      integer :: status
      character(:), allocatable :: str

      str = esmf_HConfigAsString(hconfig, keystring=keystring, index=index, _RC)
      call esmf_TimeIntervalSet(interval, str, _RC)

      _RETURN(_SUCCESS)
   end function as_timeinterval

[tclune]

Belatedly realizing that this needs to be overloaded for HConfigIter as well for consistency.

[oehmke]

I just remembered that you wanted guidance on where it should go. I think: <ESMF_DIR>src/Infrastructure/HConfig/interface/ESMF_HConfig.F90 makes sense. In there put it with the similar interfaces in alphabetical order. Regarding testing, I'm happy to add a few unit tests once it's in.

[theurich]

Catching up with ESMF... I think that these should be ESMF_TimeSet() and ESMF_TimeIntervalSet() overloads, respectively. They would live in the respective ESMF_Time and ESMF_TimeInterval source files under src/Infrastructure/TimeMgr.

[oehmke]

Yes! That makes even more sense than putting them in HConfig. Then it's more analogous with the setting of those Time classes from an ISO string.

[anntsay]

Have @oehmke talk to NASA to open a PR so the ESMF team can review and incorporate.