Some small fixes I found when updating Ariel OS to use the latest lakers:

• Some documentation on how IdCred is used would have helped, so I added an example that doubles as a doctest.
• Implementing that I found a typo (KCSS instead of KCCS); the fix is API compatible by deprecating the old version.
• The from_encoded_value constructor accepted some values that look like KIDs but are really longer. It may be a good idea to accept non-1-byte- KIDs, but this PR is designed for "all no-brainers so let's just merge it". I'm happy to provide longer KID support in there, but I don't know whether there's a limit on KID lengths in other places (won't hurt for this function to support more, though).
  [edit] Note that the way it is used internally (eg. decode_plaintext_3), the length is pre-checked to be correct due to decoder.any_as_encoded(), but that also means we're doing double work – still, a public function should do proper error handling.