Fix potential NPE from VirtualThreadPool

[lachlan-roberts]

Failure from testThreadPool – org.eclipse.jetty.util.thread.QueuedThreadPoolTest
https://jenkins.webtide.net/blue/organizations/jenkins/jetty.project/detail/PR-12181/7/tests

Error
Cannot invoke "java.util.concurrent.Semaphore.release()" because "this._semaphore" is null
Stacktrace
java.lang.NullPointerException: Cannot invoke "java.util.concurrent.Semaphore.release()" because "this._semaphore" is null
	at org.eclipse.jetty.util@12.1.0-SNAPSHOT/org.eclipse.jetty.util.thread.VirtualThreadPool.lambda$execute$0(VirtualThreadPool.java:279)
	at org.eclipse.jetty.util@12.1.0-SNAPSHOT/org.eclipse.jetty.util.thread.TrackingExecutor.lambda$execute$0(TrackingExecutor.java:47)
	at java.base/java.util.concurrent.ThreadPerTaskExecutor$TaskRunner.run(ThreadPerTaskExecutor.java:314)
	at java.base/java.lang.VirtualThread.run(VirtualThread.java:309)
Standard Error
Running org.eclipse.jetty.util.thread.ThreadFactoryTest.testThreadFactory()

Looks like a race condition where doStop() is called while an execute has been called. This PR stops semaphore from being null between the if (semaphore != null) check and the semaphore.acquire().

[joakime]

Flaky test.

testExpect100ContinueWithContentRespond413RequestEntityTooLarge(Transport)[3] H2C – org.eclipse.jetty.ee10.test.client.transport.HttpClientContinueTest

Going to merge anyway.