Add and change AbstractHomologousDiscreteCharacterData methods

[davidcerny]

This branch makes the following changes to the class AbstractHomologousDiscreteCharacterData, originally suggested by Alessio Capobianco:

(1) Add a .getInvariantSiteIndices() method, which returns the indices of all characters that are invariant, either with or without accounting for ambiguities. The code is based on the pre-existing .getNumInvariantSites() method, which has been refactored to remove redundancies.

(2) Add a .removeExcludedCharacters() method, which entirely removes excluded characters from the matrix. The idea behind this method is to do for characters what we already do for taxa, where we have both an .excludeTaxa() method (which marks one or more taxa as excluded but keeps them in the matrix) and a .removeTaxa() method (which removes them entirely).

(3) Add a test to check these methods work as intended.

(4) Use the words "exclude" and "remove" in a consistent manner by renaming the existing method .removeMissingSites() to .excludeMissingSites() (since it marks sites consisting entirely of missing data as excluded but keeps them in the matrix) and .removeRandomSites() to .replaceRandomSitesByMissingData(). This latter method neither excludes nor removes any sites; instead, it irreversibly changes their coding.

[hoehna]

Very nice with a good test!