Soil model should use biomass triplets

[jacobcook1995]

A large number of soil model pools are elemental triplets (e.g. lmwc, don, dop). It would make the links between pools easier to follow if the carbon, nitrogen and phosphorus components of each of these pools were expressed as triplets, rather than three (apparently independent) pools.

This should be fairly straightforward as the soil model already uses densities (rather than elemental ratios). The biggest hurdle will be getting the integrator to play nicely with data input that has a more complex structure