Pyvolve

Substantial update to matrix construction, specifically scaling procedure.
All nucleotide, amino-acid, empirical codon, and mechanistic codon model matrices are scaled as usual ("persite") scaling.
Heterogeneous codon model matrices are scaled such that the mean substitution rate for the _mean_ dN/dS across all categories is equal to 1.
Mutation-selection model matrices are scaled such that the mean _neutral_ substitution rate is equal to 1.

Cleaned up error messages, catching and included formal warning upon use of the 'scale_matrix' keyword argument when defining a Model.

Minor update changes the Tree class to be called Node, in order to avoid potential conflicts with the DendroPy library, which also has a Tree class. Additionally, both 'scale_matrix' and 'scaling' may be used as keyword arguments when defining a Model object to involve neutral scaling, but eventually 'scale_matrix' will be deprecated. A warning statement will now print out when this argument is given.

Neutral rate-matrix scaling should now be specified with the keyword argument <scaling="neutral"> when defining a Model object.

Addressed two issues:
1. Fixed minor bug in helpful_scripts/compute_dnds_from_mutsel.py module where the variable "frequencies" was used rather than "codon_frequencies".
2. Fixed bug in Pyvolve where the neutral scaling was improperly computed when providing fitness values to a MutSel model

Minor update to include convenience functions to return various parameters from a Model object.