* camel-sasl.c (camel_sasl_challenge): Take a GByteArray as input
	as well. Comment that you can pass %NULL for @token to get the
	initial auth data for mechanisms that are client-initiated.
	(camel_sasl_challenge_base64): Convenience function for protocols
	that use base64-encoded SASL.
	(camel_sasl_authenticated): Implement this... (it was prototyped
	already)
	(camel_sasl_new): Function to take a service name, a mechanism
	name, and a CamelService, and return a CamelSasl for it.
	(camel_sasl_authtype, camel_sasl_authtype_list): Functions to
	return CamelServiceAuthType information about SASL mechanisms, to
	allow providers to deal with them generically.

	* camel-sasl-anonymous.c, camel-sasl-plain.c: Update/simplify for
	CamelSasl changes. Both of these are single-round
	(client-initiated) mechanisms, so they don't need to keep state.
	(camel_sasl_plain_new): Removed; use camel_sasl_new instead.
	(Can't get rid of camel_sasl_anonymous_new though...)

	* camel-sasl-cram-md5.c: Update/simplify for CamelSasl changes.
	(camel_sasl_cram_md5_new): Removed; use camel_sasl_new instead.
	(cram_md5_challenge): Use md5_get_digest where possible, and
	various other minor simplifications. CRAM-MD5 only has a single
	round, so there's no need to keep track of state. This code is now
	tested (against Cyrus IMAPd) and known to work.

	* camel-sasl-kerberos4.h: Update/simplify for CamelSasl changes.
	Make only a single #ifdef HAVE_KRB4. Remove stuff from priv that
	isn't needed between rounds.
	(camel_sasl_kerberos4_new): Removed; use camel_sasl_new instead
	(krb4_challenge): Fix up the logic I broke in my previous "at
	least make it compile" fixes, update to match other changes, and
	remove IMAP-isms that shouldn't be in the generic code. This still
	isn't tested, because we're stuck behind a NAT right now...

svn path=/trunk/; revision=8462