[krbdev.mit.edu #6893] error codes from error responses can be discarded when there's e-data

[Greg Hudson via RT]

I'm going to check in a large change for this which refactors the chpw 
result parsing code and consolidates the chpw/setpw result handling.  The 
change you submitted is probably reasonable as a backstop, although it 
will not have the same behavior (since it uses the KRB-ERROR error code 
rather than the numeric code in the e-data).