Modern diag_manager:: diag_send_complete_instant has not been implemented

[uramirez8707]

Is your feature request related to a problem? Please describe.
Diag_send_complete_instant has not been implemented in the modern diag_manager, so it will do nothing and cause failures

FMS/diag_manager/diag_manager.F90

Lines 3868 to 3896 in 42f8506

	!> @brief The subroutine 'diag_send_complete_instant' allows the user to
	!! save diagnostic data on variable intervals (user defined in code logic)
	!! to the same file. The argument (time_type) will be written to the
	!! time axis correspondingly.
	!!
	!> The user is responsible for any averaging of accumulated data
	!! as this routine is not designed for instantaneous values. This routine
	!! works only for send_data calls within OpenMP regions as they are buffered
	!! until the complete signal is given.
	SUBROUTINE diag_send_complete_instant(time)
	TYPE (time_type), INTENT(in) :: time
	!--- local variables
	integer :: file, j, freq, in_num, file_num, out_num
	DO file = 1, num_files
	freq = files(file)%output_freq
	IF (freq == 0) then
	DO j = 1, files(file)%num_fields
	out_num = files(file)%fields(j)
	in_num = output_fields(out_num)%input_field
	IF ( (input_fields(in_num)%numthreads == 1) .AND.&
	& (input_fields(in_num)%active_omp_level.LE.1) ) CYCLE
	file_num = output_fields(out_num)%output_file
	CALL diag_data_out(file_num, out_num, &
	& output_fields(out_num)%buffer, time)
	END DO
	END IF
	END DO
	END SUBROUTINE diag_send_complete_instant

Describe the solution you'd like
Diag_send_complete_instant needs to be implemented as it is used in SHiELD
https://github.com/NOAA-GFDL/atmos_drivers/blob/113a25fe6fea86bb5dce8fee064740b4b96237b9/SHiELD/atmos_model.F90#L738

Describe alternatives you've considered
N/A

Additional context
N/A

[uramirez8707]

@laurenchilutti I think this is the cause of the full SHiELD test

FATAL from PE    17: atmos_4xdaily: diag_manager_mod: You skipped a time_step. Be sure that diag_send_complete is called at every time_step needed by the file.

fyi @bensonr

[bensonr]

@uramirez8707 - do you think there is a way to set the variables in the diag_table[.yaml] to get around having to implement this function?