bug fix: out of bounds array indexing error

[scrasmussen]

TYPE: bug fix

KEYWORDS: array bounds, indexing error, debugging

SOURCE: Soren Rasmussen, NSF NCAR

DESCRIPTION OF CHANGES:

• fixing out of bounds array indexing error
  • seems to be a copy-and-paste coding error. The variable index is the one being worked on, not trim(fileMeta%varNames(iTmp)) whose iTmp indexing variable isn't even being set in that do loop
• compiled with flags -O0 -g -fbacktrace -fcheck=bounds -fsanitize=address -fno-omit-frame-pointer
• before fixing the bug, running led to the following error (note I slightly modified the following message for clarity)

$ mpiexec -np 4 ./wrf_hydro
...
 Opening
At line 3368 of file src/Routing/module_NWM_io.F90
Fortran runtime error: Index '9' of dimension 1 of array 'filemeta%varnames' above upper bound of 1

Error termination. Backtrace:
#0  0x2e8e749 in __module_nwm_io_MOD_output_chrtout_grd_nwm
        at src/Routing/module_NWM_io.F90:3368
#1  0x2c98bb0 in __module_hydro_drv_MOD_hydro_out
        at src/HYDRO_drv/module_HYDRO_drv.F90:422
#2  0x2c49001 in __module_hydro_drv_MOD_hydro_ini
        at src/HYDRO_drv/module_HYDRO_drv.F90:1622
...

[rcabell]

This does seem to be producing legitimate value regressions in the Reach-Lakes case, @scrasmussen are you expecting this?

[rcabell]

While it would have been nice to have the message still have the variable name, I think this is a reasonable and safe approach

[scrasmussen]

While it would have been nice to have the message still have the variable name, I think this is a reasonable and safe approach

It seems like the message doesn't have the variable name, since the other nwmCheck calls use a variable for the variable name, but at this point they switch to using the string index in the line iret = nf90_def_var(ftn,'index',nf90_int,(/dimId(2),dimId(3)/),varid=indexVarId) call.

Thanks for reviewing! I'll go ahead and merge it