The netcdf3 routines have also been updated but not tested...

H.M.J. Barbosa:
cannot be done this way!!! idum must change between calls otherwise it is not
random there are many calls like this in this function, all with the same
problem.
i did not dare trying to fix them...

The read statement didn't expect the same formatting as in surfdata.t
This updates fixes that problem and also adds a test to make sure the table
is read correctly.

Receptor data for netcdf is currently written by the standard netcdf output
routine (write_ncconc.f90). Therefore, no receptor data should be written by
the concoutput_* routines when netcdf output is selected.
Conflicts:
	concoutput_reg.f90

Conflicts:
	write_ncconc.f90

concoutput_irreg.f90 didn't require any changes.
Conflicts:
	concoutput_reg.f90

The model will now be tested, if I'm lucky, everything works as it should.
Landuse will not be interpolated between time step because:
1. seasonal changes are instant, so keeping track of snowmelt on a higher
	frequency than input data is read (hourly?) makes little sense.
2. Other parameters are interpolated instead, such as U*, vd etc.
	(how true/effective is this?)

Should now follow FLEXPART coding standard.

Also indented block to follow FLEXPART coding standard (didn't touch any other
part of the file, so these are still indented wrong).