optpart 1.1-0

There are several interesting changes to 1.1-0.

1) there are now four different iterative re-assignment classification
algorithms:

   1.1) optpart - to optimize the partana statistic
   1.2) optsil - to optimize silhouette widths
   1.3) optindval - to optimize Dufrene and Legendre indicator values
   1.4) opttdev - to optimize table deviance

2) the iterative re-assigmnent algorithms now all have multiple inheritance.
   2.1) their own specific class, and 
   2.2) clustering - which is shared with several other simpler algorithms


