FMS 2 I/O logic errors in read_climate_nudge_data_init

[menzel-gfdl]

Describe the bug
There is an indexing error here:

GFDL_atmos_cubed_sphere/GFDL_tools/read_climate_nudge_data.F90

Line 156 in b67be1d

if (trim(Files(n)%axes(j)) .eq. trim(required_axis_names(j))) then

As written, this line assumes that the dimensions in input file appear in the same order as the required_axis_names. This in general will not be true. Also, this code does not store the time axis data correctly. How exactly does the code determine the time axis? Is it required to be the only unlimited dimension?

To Reproduce
Attempt to run a "nudged" experiment. A few different fatal or fortran errors are possible. Some come from here:

GFDL_atmos_cubed_sphere/GFDL_tools/read_climate_nudge_data.F90

Lines 572 to 590 in b67be1d

	subroutine check_resolution (axis_len)
	integer, intent(in) :: axis_len(:)
	if (size(axis_len(:)) .lt. 2) then
	call error_mesg ('read_climate_nudge_data_mod', 'incorrect number of array dimensions', FATAL)
	endif
	if (axis_len(1) .ne. global_axis_size(INDEX_LON)) then
	call error_mesg ('read_climate_nudge_data_mod', 'incorrect array dimension one', FATAL)
	endif
	if (axis_len(2) .ne. global_axis_size(INDEX_LAT)) then
	call error_mesg ('read_climate_nudge_data_mod', 'incorrect array dimension two', FATAL)
	endif
	if (size(axis_len(:)) .gt. 3) then
	if (axis_len(3) .ne. global_axis_size(INDEX_LEV)) then
	call error_mesg ('read_climate_nudge_data_mod', 'incorrect array dimension three', FATAL)
	endif
	endif
	end subroutine check_resolution

I have also seen intrinsic fortran errors in the get_cal_time function from FMS, because if the wrong axis is identified as the time axis, the units may have less than 10 - 15 characters causing a buffer overflow.

Expected behavior
The routine should correctly identify the time, lon, lat and lev axes.

System Environment
This was first seen using the intel 18 compiler on the Gaea hpc system.

Additional context
I believe I have a fix, and will create a pull request.

[bensonr]

@menzel-gfdl - thanks for creating a fix/patch. This must be a use case for which we do not have a test within the RTS system for GFDL models at this point in time.

[laurenchilutti]

Fixed by PR #138