[R] predicted survival curves from a Cox time dep model

[Terry Therneau]

Calculating predicted survival curves in the case of a time dependent variable 
is a thorny problem; not computationally but conceptually.  It is not always 
clear exactly what you WANT to plot. My book has a whole section on the subject, 
which is only to say that it is too long a subject to summarize in an email 
response.  Also see 
   Simon and Makuch, Statistics in Medicine 3:35-44, 1984
   Klein, Keiding and Copelan, Statistics in Medicine 13:2315-2332, 1994
   
which shows that the debate is an old one.  There are many other good papers 
which I haven't listed.

  Issue 2: the cluster() directive confuses the survfit function.  This is a 
bug.  I will address it.
  
  	Terry Therneau