Merge bind-simmat branch to master: adding new function smbind

[psolymos]

The new function smbind is related to issue #164 works as expected. Examples cover the use cases and results are as expected. Also fixes some issues with start/end/thin values in as.ts.oecosimu, simulate.nullmodel, and update.nullmodel.

[jarioksa]

I have run some tests stratifying species by taxonomy, and everything seems to work fine. The usage requires some familiarity with R intricacies, though. Perhaps permatswap or permatfull interface could be used to make stratifying easier.

So far my only comment is that you should not reset RNG in the user session. This would happen with example(nullmodel). I have suggested a change to this in daad496. This is similar as in man/orditorp.Rd, and it would be still better to save the state of RNG and reset to it on.exit() (in both cases).

It is a question of taste, but I do prefer to have real data sets in examples instead of simulated ones.

[psolymos]

I have no problem with changing the matrix to data(dune); x <- as.matrix(mite) or something similar in example(nullmodel). That would address both the seed reset issue and preference towards real data sets.

I merge this branch and follow up with the example and null model error messages as references in #177