Add the start year as variable to the template to capture future climate
scenarios. In addition, update the naming scheme that maps run IDs to
climate scenarios.

The script can either create a specific run scripts for a given
experiment ID, or all run scripts for the Monsoon 2.0 project.

Although the run easiliy finishes wiithin the time limit, it is also a
nice to test if the model restart works as expected.

1. Increase output frequency for 2D and 3D fields significantly
2. Adapt upper limit for tracer output
3. Do not store grids in 2D fields. This seems to cause crashes when
   restarting the model.

* Decrease the vertical level at which the substepping starts.  This
  fixes problems in strong wind situations around Antarctica.
* Enable advection substepping for cloud ice and cloud water.

* Decrease the bottom threshold for substepping. This (hopefully) fixes
  runs with very large winds around Antarctica.
* Enable substepping for cloud-ice advection

Currently, there is an issue with ecCodes and the Nvidia compiler. As a
workaround, use netCDF output.

Also update to the latest configuration file for JSC.

The set timelimit of 24h includes a safety margin as three months are
only expected to run 18 hours. But a division into 2*3 months seems to
be the safe choice.

We have to double check, if they are acutally required