[pyicon_dwd] Use CERES and GPM data as reference for atmospheric plots.

    More details:

    *) CERES data are used as reference for
       -- TOA radiation fluxes
       -- total cloud cover

    *) GPM data are used as reference for
       -- total precipitation
       -- note, that for P-E plots, still ERA5 is used as reference,
          as evaporation is available only from ERA5

    *) If any of the 2 datasets (CERES, GPM) does not exist or not
       defined, the fall-back option is to use ERA5 for the above
       variables.

    *) Tools are added to prepare necessary CERES/GPM reference
       data sets:
       tools/prep_ceres_pyicon.py
       tools/prep_gpm_pyicon.py
       Thay are written in xarray and does the following:
       -- interpolates the data to arbitrary lat-lon
          resolution
       -- calculates time-mean maps for arbitrary time period
          for 2D fields
       -- calculates global-mean (cell-area weighted) times
          series

    *) pyicon/quickplots/qp_driver.py is updated so that CERES/GPM
       data are used as reference for bias & RMSE plots for the
       relevant variables (see above).

    *) pyicon/quickplots/qp_compare_atm_dwd.py is updated so that
       CERES/GPM data are used as reference for bias & RMSE plots
       for the relevant variables (see above).

    *) pyicon/quickplots/pyicon_quickplots.py is updated in order
       to show CERES/GPM data also in times series plots for the
       atmosphere. The option 'if use_tave_int_for_ts:' has been
       also rewritten wrt the slicing of the time series data
       given by --tave_int time bounds.

    *) tools/run_qp_driver_*_dwd.sh are updated by the path and
       file name of the CERES/GPM reference data.

    *) tools/run_qp_compare_*_dwd.sh are updated by the path and
       file name of the CERES/GPM reference data.