[FEATURE REQUEST] Save KPP global variables into a derived type object

[yantosca]

This would be for much later down the road. Right now KPP global variables are referenced by F90 USE statements. But there may be some model implementations (such as in WRF) where several chemistry domains get solved simultaneously. For such a case like that, it would be preferable to have all KPP-generated module variables stored in a F90 derived type object so that the model could instantiate as many instances of the object as there are chemistry domains being solved simulataneously.

This would be a "nice to have" but is not a high priority. I just wanted to list this here so that we don't lose sight of it.

[jimmielin]

Looks like this feature will be useful for MATLAB as well (see #45) as suggested by @JPSTUTZ.

We can move towards using structure arrays for MATLAB, which is the equivalent for derived types.