Improve 'none' algorithm handling

[danleyden]

when decoding a valid token with algorithm: none, unless explicitly
requesting not to verify the token, it could not be validated and raised
DecodeError with a message about insufficient segments.

This error message is misleading, because there are the correct number
of segments provided for that algorithm.

With this change, when a token with algorithm: none is provided:

• if the caller requests verification && does not specify none as an
  algorithm (default behaviour) it now raises IncorrectAlgorithm,
  which is a subclass of DecodeError to make the issue more clear.
  This is technically a minor change, but should not be breaking -
  it is returning a subclass, so the same rescues will still work. The
  message provided will be different.

• if the caller requests verification && specifies none as an allowed
  algorithm, it verifies the claims and decodes the token as it would
  for a valid, signed token.
  This is new behaviour supporting claims verification for 'none' which
  was not previously available and is only "accessed" through explicit
  settings

• if the caller explicitly requests no verification, the token is
  decoded without checking anything (no change in behaviour)

[sourcelevel-bot]

Hello, @danleyden! This is your first Pull Request that will be reviewed by SourceLevel, an automatic Code Review service. It will leave comments on this diff with potential issues and style violations found in the code as you push new commits. You can also see all the issues found on this Pull Request on its review page. Please check our documentation for more information.

[anakinj]

This looks good to me