[krbdev.mit.edu #2644] fakeka.c should use life_to_time/time_to_life

[Tom Yu via RT]

fakeka.c hardcodes its own table of "cmu_seconds" which is slightly different from the one in 
v4lifetime.c.  The table hardcoded in fakeka.c appears to come from the old krb524d code, 
which is known to differ from the table in the AFS source code.  The table in v4lifetime.c is 
known to be identical with the one in the AFS source code.  (The differences are probably due 
to roundoff errors from using a rounded value of the ratio between successive values rather 
than computing the exponential curvefit between the fixed points and using maximum 
precision prior to final rounding.)