To do (as recently discussed with Jose and Frank)
high priority
- allow for additional covariates
- more robust fitting procedure (use parameter 
  bounds for non-linear parameters)

lower priority
- allow for dose estimation with active control group
- bootstrap methods for precision of the target dose
- allow for different maxEffs (and baselines)
  for planning
- allow for different clinical relevance thresholds for
  analysis
- add covMu argument for planning functions
- Function to evaluate power and precision of the dose
  estimate (under a certain dose design) by simulation
- Function to interactively choose guesstimates
- verbose argument for print.summary.MCPMod?
- function to calculate MED-optimal robust designs