[krbdev.mit.edu #5839] fix krb5_string_to_keysalts() string processing infinite loop and default keysalt type

[Ken Raeburn via RT]

The first part looks like a no-brainer, go ahead...

I'm guessing the second part changes the software behavior from requiring the salt type to be 
explicitly given to defaulting to normal salt type?  I think it's probably okay, but I could imagine 
someone might think having the salt type explicit is a good thing, maybe.  And if it is a 
behavior change, there may be doc changes needed as well.