Comment-driven query timeouts do not work when shard-targeting

[aquarapid]

Tested on a current master release (c134e38).

• Connect to a Vitess instance. Using the 101_initial_cluster.sh local examples in my case.
• Run a select sleep(1) query with a small timeout (1 ms). See that it gets timed out.
• Target a specific shard, and run the same query again. Observe that it does not get timed out.

$ mysql -c -u root -P 15306 -A
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 7
Server version: 5.7.9-Vitess Percona Server (GPL), Release 34, Revision 2e68637
.
.
.
mysql> use commerce;
Database changed
mysql> select /*vt+ QUERY_TIMEOUT_MS=1 */ sleep(1);
ERROR 1317 (HY000): vtgate: http://localhost:15001/: target: commerce.0.master, used tablet: zone1-100 (localhost): vttablet: rpc error: code = DeadlineExceeded desc = context deadline exceeded
mysql> use commerce/0;
Database changed
mysql> select /*vt+ QUERY_TIMEOUT_MS=1 */ sleep(1);
+----------+
| sleep(1) |
+----------+
|        0 |
+----------+
1 row in set (1.00 sec)