meta-model: multiple LDA_TS structure: function and options list

[juniperlsimonis]

the next step upwards in the model structural hierarchy will combine multiple individual full runs of an LDA_TS model

we need to determine what we'll call it (mLDA_TS, LDA_TSs, ??), but this function will allow for multiple runs of the same model front-to-back but with perhaps different data, for example with leaving-out-cross validation approaches that work with evaluation of the full model at the end, rather than at each component.

this functionality will also allow for a more active decision making about each of the component models (see #146 #147 #148), allowing the user to for example decide on which topic model or time series model to use.

accordingly, we'll need a higher-level options list that controls the full model run, evaluation, and selection.