Fix 100% CPU usage with DependentAdminClient (confluentinc#1949)

* Add default value of cancellationDelayMaxMs for DependantAdminClient

* Update CHANGELOG

* Use ExtractCancellationDelayMaxMs instead of hardcoding

Author: anchitj

CHANGELOG.md

@@ -10,6 +10,7 @@
 ## Fixes
 
 - During a group rebalance, partitions are now always revoked as a side effect of a call to Consume, whether or not a partitions revoked handler has been specified. Previously, if no handler was specified, the timing of when the consumer lost ownership of partitions during a rebalance was arbitrarily, frequently resulting in an erroneous state exception when committing or storing offsets.
+- Fixed 100% CPU usage with `DependentAdminClientBuilder`. 
 
 
 # 1.9.4

src/Confluent.Kafka/AdminClient.cs

@@ -768,7 +768,8 @@ private SafeKafkaHandle kafkaHandle
         ///     or AdminClient handle.
         /// </param>
         internal AdminClient(Handle handle)
-        {                            
+        {
+            Config.ExtractCancellationDelayMaxMs(new AdminClientConfig(), out this.cancellationDelayMaxMs);                          
             this.ownedClient = null;
             this.handle = handle;
             Init();