* Compare with Jim Lindsey's "stable" package
  --> http://www.commanster.eu/rcode.html
  Maybe even ask him to make it CRANny
  (maybe after transfering maintainership ?)

  At SfS ETH: source in /usr/local/app/R/R_local/JimLindsey/stable/
  installed in          /usr/local/app/R/R_local/JimLindsey/library

* stableMode():  [MM] I'm pretty sure that the  'beta.max' argument is
  		 unneeded and could be eliminated.

* parametrizations -- pm=0, 1, 2 ---
  provide functions for the parameter transformation that go from
  other to pm=0 -- and back
  and then show how to, or  use these to to provide transformations
  from all pm's to any other.

* [qp] check the  lower.tail & log.p argument settings
  NB: This check *should* happen in  fBasics  'checkdist' ...

* curve(dstable(exp(x), alpha= 0.1, beta=1, pm=1, log=TRUE), -15, 0)
  very much depends on  zeta.tol ... that should somehow depend on alpha or zeta;
  --> tests/dstab-ex.R ("beta = 1")

* As easy alternative to  fBasics::stableFit() define one  based
  on stats4::mle (and possibly also a  quantile-matching alternative).
