optimize MLE

MLE models are much slower now than they could be with faster R implementation or, maybe better, Rcpp. There's also some funkiness with `nlm`+`ode`+`lsoda` that might possibly be solved with `optim` or `nlminb` (haven't tried)
- [ ] if i'm going to optimize for time, i should probably start tracking runtime now
- [ ] implement core loops in Rcpp a la https://github.com/GLEON/LakeMetabolizer/blob/master/src/mleLoop.c and http://adv-r.had.co.nz/Rcpp.html
- [ ] consider nlminb (with constraints?) rather than nlm? http://stats.stackexchange.com/questions/9535/when-should-i-not-use-rs-nlm-function-for-mle, http://stackoverflow.com/questions/3757321/moving-beyond-rs-optim-function