Recalculate maxQueueSize based on shards count

Currently maxQueueSize is 256 by default.
Problem is that some hosts can have 1 shard, others can have 256 shards.
While having cap of 256 pending requests for 1 shard host is ok.
For host of 256 shards it is way to low.
This commit scales maxQueueSize by number of shards, which allow default
value works for hosts of any size.

Author: dkropachev

driver-core/src/main/java/com/datastax/driver/core/HostConnectionPool.java

@@ -514,6 +514,7 @@ ListenableFuture<Connection> borrowConnection(
 
     int shardId = 0;
     if (host.getShardingInfo() != null) {
+      maxQueueSize = host.getShardingInfo().getShardsCount() * maxQueueSize;
       if (routingKey != null) {
         Metadata metadata = manager.cluster.getMetadata();
         Token t = metadata.newToken(partitioner, routingKey);

driver-core/src/main/java/com/datastax/driver/core/PoolingOptions.java

@@ -462,7 +462,8 @@ public PoolingOptions setPoolTimeoutMillis(int poolTimeoutMillis) {
   }
 
   /**
-   * Returns the maximum number of requests that get enqueued if no connection is available.
+   * Returns the maximum number of requests per shard that get enqueued if no connection is
+   * available.
    *
    * @return the maximum queue size.
    */