#450 robustness of mossco_time.F90 with respect to string for time units

[Markus Kreus]

My MOSSCO simulation crashed because of a badly formatted timestring (date as "2002-1-1" instead of "2002-01-01") given for "time:units" in a netcdf file which was created for restart by means of the CDO operator.

Even though this bug was reported to CDO developpers ([CDO - Bug #7902] (New) badly formatted timestring generated by setreftime ), I would appreciate a more informative error handling as MOSSCO logfiles did not provide much information to tackle the core of the problem.

Cheers, Markus