[nrf fromtree] Bluetooth: CSIP: Fix off-by-one in in lock restore

If the lock request was rejected by a set member we should
restore any previously written logs in reverse order.

However there was a off-by-one error in
csip_set_coordinator_write_lock_cb which caused us to attempt to
release member[1] instead of member[0] if member[1] was the
one that rejected the lock request.

Additionally, the lock_set_complete would be called
prematurely before we get the response from the restore
request.

Signed-off-by: Emil Gydesen <[email protected]>
(cherry picked from commit 7c40b07)

Author: Thalley

subsys/bluetooth/audio/csip_set_coordinator.c

@@ -1114,7 +1114,7 @@ static void csip_set_coordinator_write_lock_cb(struct bt_conn *conn,
 
 			active.members_restored = 0;
 
-			member = active.members[active.members_handled - active.members_restored];
+			member = active.members[active.members_handled - 1];
 			client->cur_inst = lookup_instance_by_set_info(member, &active.info);
 			if (client->cur_inst == NULL) {
 				LOG_DBG("Failed to lookup instance by set_info");
@@ -1131,10 +1131,10 @@ static void csip_set_coordinator_write_lock_cb(struct bt_conn *conn,
 				active_members_reset();
 				return;
 			}
+		} else {
+			lock_set_complete(err);
 		}
 
-		lock_set_complete(err);
-
 		return;
 	}