Add functionality to do inverse prediction

[psolymos]

Create new branch (inverse-prediction) and implement the following functions:

• ipredict generic and methods (default, opticut): currently called calibrate, but ipredict should better reflect the intention. This will only run a single instance with any ynew/xnew arguments.
• multiclass measures based on Sokolova et al.
• calibrate method to implement LOO. Implement comparison based on multiclass measures similarly to AIC but meybe only for 2 objects: LOTO (leave one taxon out) vs. reference (all taxa LOO)

[psolymos]

Write tmpfile for model directly to avoid multiple thread clashes.

[psolymos]

Deal with possibly different species subsets in object and ynew:

• species present in ynew but not in object: drop
• species present in object but not in ynew: 0 value is information, thus setting it to 0 is not correct. Need to subset JAGS input data to contain the intersect of species.

This is best achieved by a subset method for the opticut/uncertainty model classes: carefully subset species (modifying other attributes etc. as needed). This is useful and safe option.

[psolymos]

• [x] dcoptions used to avoid overwriting model file when running in parallel -- needs testing
• [x] subset method added and used in ipredict.opticut
• [x] add multiclass functions
• [x] implement a multinomial fitting function (multicut) modeled after opticut and optilevels (all dist, because it is already used in rankComb)
• [x] modify ipredict.default for multicut objects (binomial and poisson cases only).
• [ ] write documentation and add LOSO/LOTO examples in \dontrun or implement fast version:
  • leave-one-site-out (LOSO): opticut((n-1) x m), ipredict(1 x m), repeat n times, multiclass assessment
  • leave-one-taxon-out (TOTO): opticut(n x (m-1)), ipredict(n x (m-1)), repeat m times, assess
• [x] predict method for opticut and multicut objects (with args: object, xnew, gnew; all dist types)
• [x] predict method for multicut objects (with args: object, xnew, gnew; all dist types)
• [x] MCMC and analytic options for ipredict: analytic should implement additive loglik (binomial and poisson cases only)

[psolymos]

Needed methods for multicut class:

• [x] print,
• [x] summary,
• [x] print.summary.multicut,
• [x] plot,
• [x] fitted,
• [x] predict,
• [x] ipredict,
• [x] getMLE,
• [x] strata,
• [x] subset,
• [x] as.data.frame,
• [x] as.data.frame.summary.multicut.

Not sure if uncertainty and related methods need to be implemented as well. It is not needed for ipredict, stratification is fixed, and bootstrap is kind of trivial in that case.

Need to test all the dist cases (compare coef with getMLE output)

[psolymos]

Most of the prep work done for milestone v1.0. Here is the list of remaining items for milestone v1.1:

• [x] ipredict methods for opticut and multicut objects implementing IP.
• [x] add multiclass functions for assessing IP (branch multiclass).
• [x] implement LOSO/LOTO examples and functions for analytic and MCMC based IP (call it ipower referring to [posterior] indicator power).

[psolymos]

Need to figure out final object structures, print/summary/plot methods, write documentation and tests. Otherwise the multiclass branch has the new functionality in /extras.

[psolymos]

kappa stat needs to be fully fledged out, also considering prior weights.