Add `guppy fpd`

[cmccoy]

Merge wpd, pd, and entropy into a single subcommand: guppy fpd

guppy fpd should be similar to guppy entropy, with a few modifications:

• add a pd column with phylogenetic diversity
• add a wpd column with fully-weighted phylogenetic diversity (kappa=1)
• add a flag --kappas, which accepts a delimited list of values for the exponent in wpd, and causes corresponding columns wpd_{kappa1}, wpd_{kappa2}... to be included in the output.

[matsen]

Sounds good. I like the use of --exponents. Let's try to avoid the word "alpha" for the exponent parameter as much as possible, and certainly in the documentation. "exponenent" is the appropriate replacement.

In addition, it would also be nice to have a --pendant flag. The idea is to make the same calculations as if the masses were attached via pendant branch lengths.

This doesn't require a tree traversal, so I think it's best done in a second pass through the placements. Simply sum the pendant branch length times the transform of the fractional mass for each placement. Transform here is the entropy function, quadratic entropy, etc.

[matsen]

A paragraph for docs:

Warning: using the --pendant branch length flag is not the same as running the equivalent analysis with a fully resolved tree. For example, two sequences that are sister in the fully resolved tree should get placed next to each other, but their common ancestry will not get factored into the analysis.

[matsen]

I changed my mind about the --exponent thing. I think that we should call the exponent kappa and should use the --kappa flag.