[FIX] generate_tracking_message_id: use more randomness to avoid duplicates

[FIX] generate_tracking_message_id: use more randomness to avoid duplicates

When several message IDs had to be generated for the same
record at the same time, there was a high chance to
get non-unique results. This possibly lead to data loss because
some mail implementations might ignore multiple mails with the
same ID (including OpenERP itself). On most operating systems
the available time resolution precision is better than what
we used (due to the float rounding of `%s`).
Adding a bit of randomness doesn't hurt, as OpenERP will
be used in increasingly distributed environments, so
we now add an extra random part as well.

bzr revid: odo@openerp.com-20121022162421-qr7nq8idihp5781u