s4:torture/smb2: fix and improve the smb2.lock.replay test

s4:torture/smb2: fix and improve the smb2.lock.replay test

The test was wrong in that it used an invalid
lock sequence bucket (65) where it actually wanted
to use a valid on (64), and hence the test results
(which were adapted to the real responses) were not
quite logical.

This patch fixes this and also improves some of
the comments so that the flow of the patch becomes
a little more obvious.

Pair-Programmed-With: Günther Deschner <gd@samba.org>

Signed-off-by: Michael Adam <obnox@samba.org>
Signed-off-by: Günther Deschner <gd@samba.org>
Reviewed-by: Stefan Metzmacher <metze@samba.org>