c   To Do:
c   ------
c
c   Get rid of *clim* files and absorb the code in the non-clim
c   equivalents.  There is no reason to distinguish climatological
c   variables from others input fields.  VAriable repeatPeriod
c   makes ALLOW_CLIM_CYCLIC obsolete.
c
c   Many assumptions that are made in the current version should, and
c   can, be relaxed in later versions:
c
c   - equidistancy in time -->  general data set (non-equidistant in
c                                                 time)
c
c   - climatologies        -->  arbitrary subinterval of a year, not
c                               just monthly values.
c
c   - relaxation           -->  relaxation could be done to a given
c                               data set over the model's integration
c                               time. This would generalise the clima-
c                               tology block. One could have 'cyclic'
c                               and 'non-cyclic' fields. In the first
c                               case, the cycle length has to be spe-
c                               cified ('monthly','yearly','period') as
c                               well as the start of the cycle ( irec=1
c                               corresponds to a certain calendar date).
c                               The second case can be treated analog-
c                               ously.