Bootstrapping!

[mingrisch]

this should be implemented as a method in the CompartmentModel class. Potential difficulties include:

• conversion from raw to physiological parameters: This needs to be done for each fit - before distribution statistics are determined. Essentially, we need to call the (possibly renamed, see #19) get_parameter method and store the outcome in the same array as the raw bootstrapping estimates.
• How do we handle the results? * have the parameter dictionary contain tuples for each parameter, with (lower estimate, median/mean, upper estimate)? This would be elegant, and it would not break any code that does not bootstrap. * introduce a second dictionary, with the error estimates?
• Wouldn't it be nice to have the bootstrap array available? Visualize the distributions, maybe?

[mingrisch]

Looks like a job for @AndreasMittermeier :)

[mingrisch]

This issue is, in fact, closely related to #19 I am not happy with the current implementation of the bootstrapping routine, it is somewhat cumbersome to implement the conversion from fit parameters to physiological parameters twice - in the same function!

Instead, this conversion should be done by a possibly internal function - see #19 which should be called after the first fit and also in each bootstrap iteration.