Mehmet Cüneyd Demirel authored Oct 02, 2020 and Sebastian Müller committed Oct 02, 2020

Global field capacity dependency to root fraction coefficient based on Demirel and Jarvis/Feddes for evapotranspiration reduction (processCase(3)=3/4) - implemented by GEUS and ITU

Maren Kaluza authored Nov 17, 2019 and Stephan Thober committed Nov 17, 2019

ToDo:

- rename observed tws to twsa
- divide by number of cells, domain wise

the variable name now also is part of the datastructure and set in the
mHM read config. With this change, and since the nTimeSteps can be
handled individually for each domain and isn't part of the reading
anymore, the reading of optional gridded data works for all
et/tws/sm/neutrons in the same way.

also deleted create_dummy_timeseries since it is not needed with the new
way to read tws

L1_tws is loaded in mHM_driver and passed to the reading function. This
is a bit different than the current style but helps to get a better
overview.

tws already uses it, others can be easily adapted
it contains:
- the data
- a mask (probably should have missing value?)
- the specific timestep (now with the oportunity to use a separate one
  per domain)
- the directory

ToDo:
- adapting objective function
- removing the old way
- adapting the checkcases
- adapting the namelist

- added objective function 32 in mhm_driver
- apply mask to boxcox transformation for observed streamflow

- domainMeta now has a new member: doRouting
    * in case of subdomains routing is now only performed if
      doRouting(iDomain) is .true.
- the new opti_function now calls mhm several times for a different
  set of domains, each time with the corresponding opti data only

opti_domain_indices is an index list of domains used in the mhm call. It
is an optional parameter and is passed within the opti_functions.

case 32 is added to the opti_functions.

also, new communicator

optimize is only knwon from common_mRM_mHM and therefore cannot be used
for the distribution of domains to the MPI processes already.
In case of sequential run, the master process nbow has nothing to do.

split reading of configuration and reading data of mRM so parts of it can be done in the configuration part

this resolves the incorrect estimation of daily discharge

there has to be a boolian in the dds and others that determines how
often the objective functions should be called by the subprocesses

- fixed some bugs in formatting of files

- finalized beta version of format_doxygen_docs_in_f90.py
- fixed remaining update in documentations

         -reformatted many lines

- finalized script and rewrote doxygen strings for all routines except
  in lib (pre-interface strings not yet supported) 
- remaining failing test_cases (NAG will provide patch):
  - NAG fails with open-mp false during read-in of namelist
- still TODO: check write optimization restart file