Support extra sib in twin models

[tbates]

Following changes in #160, umx's twin models can reasonably easily be extended to include nSib >2 (need to change the expCovMZ/DZ algebras, the extra means should all be done automagically.

• umxACEv
• umxACE

[mcneale]

This is very cool! The workshop involved exactly this example yesterday, but done with OpenMx. Sadly the twin stuff has been shoe-horned into one 2-hour practical session plus quite a few background vids.

[tbates]

H asked me about it and it turned out to be a fun problem to code into 'umx'.

Was using data to test where a covariate "age" correlated .8 with the DV "bmi" in twins and zero in the third sib, which made it look like there was an error in the new functions (h2 estimated at 0 in the 3 sib+cov model)...

Handily, the ACEv model however also gave the largest negative Variance estimate I've ever seen, which was a nice heads-up that the data might not have been simulated correctly!

So +1 for ACEv :-)