today I've updated some authentication methods (ldap, db, shibolet...) in order to detect if they are working under one UTF-8 Moodle server to avoid the urtf8_decode() they were executing by default. With this, exactly the same behaviour than in 1.5 is expected, always assuming that original data is UTF-8.
What I wanted to propose is the creation of one new field under each external authentication method, to be able to specify the encoding of the retrieved data. With this we could work against any DB, server... no matter of its encoding because we could convert easily to UTF-8 before inserting data in Moodle.
Ideas will be welcome, ciao
P.S.: For your info, this is point 34 of http://docs.moodle.org/en/UTF-8_scripts