To do:
- allow for dose finding with active control  
- additional methods for bootMCPMod objects (plot, summary method)
- calculation of lower bound on design criterion (for validating numerically
  found optima)
- Function to interactively choose guesstimates
- Add Bayesian (and bootstrap?) fitting capabilities in gFitDRModel?
- Add a way to model dose-time-response data (i.e. add a way of handling
  missing data) for commonly encountered endpoints (e.g. normal, binary, 
  time-to-event)?
- Extend calcOptDesign to non-normal data