Improvements for Mixed Effects Models

[jsilve24]

• Computational Efficiency. This is my number one priority but also the least clear of how to accomplish.
• Currently no option/control parameters are passed to lmer.
• Currently uses lme4 which cannot handle more complex correlation structures.

This builds on #14 from @kyle-mcgovern