Caches executor wrappers (#2033)

without this change whenever a traced executor is created (either via a proxy or directly) we create a new instance of that traced wrapper. The problem with that is such that we create a lot of objects (e.g. each getExecutor() would return a new instance of the traced version). There are projects such as Micrometer that create WeakReferences to those objects. That means that if we don't cache the traced instance then we will lose the WeakReference upon the first GC.
with this change we're introducing new static wrapper methods that cache the traced instance for a given delegate. That means that not only will we create fewer objects but also we will honour the WeakReference mechanisms.

fixes gh-2020

tests/common/src/main/java/org/springframework/cloud/sleuth/instrument/async/LazyTraceThreadPoolTaskSchedulerTests.java

@@ -20,6 +20,7 @@ import java.time.Duration;
 import java.time.Instant;
 import java.util.Date;
 import java.util.concurrent.Callable;
+import java.util.concurrent.Executors;
 import java.util.concurrent.RejectedExecutionHandler;
 import java.util.concurrent.ThreadFactory;
 
@@ -94,6 +95,8 @@ public abstract class LazyTraceThreadPoolTaskSchedulerTests implements TestTraci
 
 	@Test
 	public void getScheduledExecutor() {
+		BDDMockito.given(this.delegate.getScheduledExecutor()).willReturn(Executors.newScheduledThreadPool(1));
+
 		this.executor.getScheduledExecutor();
 
 		BDDMockito.then(this.delegate).should().getScheduledExecutor();