Iterate on `fit_species` about allometric

[brycefrank]

I think the cleanest thing to do is to enforce some structure here. 90% of cases will be genus-species pairs. That is sufficient to me. The other 10% are higher-order taxonomic groups (fine, as long as they are defined) or vague common-named groups (e.g., "cottonwood" in the Brackett report). The latter case is tougher to handle but I think fairly rare (maybe ~1%, usually the genus-species pair can be deduced, this one was an exception).

So, for a given allometric model we should know: what is the taxon. The taxon will be defined as the deepest defined group in the taxonomic hierarchy for the model, starting probably at Family (I can't imagine we need orders...). For example, a douglas fir model goes all the way down to Pinaceae Pseudotsuga menziesii. Some models might not have one (e.g., the Behre's hyperbola), and that is ok. This information should be stored at the AllometricModel level. Models can also have multiple taxons.

This information should be stored as part of some type of metadata tracking the "fit" of the model. This is more general than the taxon and might include things like age class, region, etc. Brackett is a good example (interior vs coastal). Similar to the variable naming, some standard verbage for commonly used data of this kind will be good.

[brycefrank]

These are now called descriptors, AllometricModel has an optional slot @descriptors which contains a list.

Here is an example:
list(country = 'US', genus = 'Pinus', species = 'ponderosa')