A test in src/mailman/model/docs/registration.rst fails if Python default encoding is UTF-8

The Invalid email addresses section in the doctests in src/mailman/model/docs/registration.rst attempts to register various invalid email addresses and checks that they are rejected.

In particular, the test

    >>> registrar.register(mlist, '\<email address hidden>')
    Traceback (most recent call last):
    ...
    InvalidEmailAddressError: \<email address hidden>

will fail if python's sys.getdefaultencoding() returns 'UTF-8' because the invalid address is reported with the UTF-8 encoding of nonbreaking space rather than \xa0.