Potential problem in histwrite2d_xios for non-IR/VIS outputs

[emillour]

Routine histwrite2d_xios has been adapted (r2735) to be able to output IR of VIS outputs (comparable to writespecIR and writespecVI) but the piece of code:

    if ((size(field) .eq. L_NSPECTI) .or. (size(field) .eq. L_NSPECTV)) then 
!$OMP MASTER 
      ! only for spectral stuff: IR_Bandwidth and VI_Bandwidth
      call xios_send_field(field_name,field)
!$OMP END MASTER
      return
    endif

will cause problems if L_NSPECTI or L_NSPECTV (number of IR or VIS spectral intervals) happens to be the same as that of a regular 2D (surface) field's "klon" value... not too likely, but it can happen...

To be improved? Maybe with added checks to the corresponding "klon_glo" and/or ... ?
Or at least give a clear warning of a possible collision if L_NSPECTI==klon or L_NSPECTV==klon