nlip: Negative Log Incomplete Posterior
In NathanWycoff/iPLSV:

The log posterior with the latent class assignments integrated out, evaluated at PHI, THETA, PSI as specified, given docs.

1	nlip(PHI, THETA, PSI, docs, eta, gamma, beta, soft_PHI)

`PHI`	A K by V matrix with simplex valued rows giving the the probabilty of words (cols) in topics (rows).
`THETA`	M by P real valued matrix, giving P-d locations of documents.
`PSI`	K by P real valued matrix, giving P-d locations of topics
`docs`	A term frequency matrix, that is, one with a row for each document, a column for each vocab word, and integer entries indicating the occurence of a word in a doc.
`eta`	The exchangible dirichlet prior on words in a topic.
`beta`	The precision for topic locations, a positive scalar.
`soft_PHI`	A boolean, if TRUE, PHI was provided on the logodds scales.
`gama`	The precision for document locations, a positive scalar.