Introduce Dispatchers.shutdown

[qwwdfsad]

Fixes #2558

[dkhalanskyjb]

The comments I made in #2903 (review) are somewhat applicable here: I think that we need to establish and document some specific behavior for cases where there are some pending jobs at the moment of shutting down the dispatchers.

As is, I managed to get a highly confusing and uninformative error with this code:

        val j = GlobalScope.launch {
            delay(1000)
            println("won't be reached")
        }
        GlobalScope.launch {
            Dispatchers.shutdown()
            j.join()
        }

The error:

Exception in thread Thread[DefaultDispatcher-worker-1 @coroutine#1,5,main]: java.lang.AssertionError
java.lang.AssertionError
	at kotlinx.coroutines.EventLoopImplPlatform.reschedule(EventLoop.kt:17)
	at kotlinx.coroutines.EventLoopImplBase.schedule(EventLoop.common.kt:362)
	at kotlinx.coroutines.EventLoopImplBase.scheduleResumeAfterDelay(EventLoop.common.kt:235)
	at kotlinx.coroutines.DelayKt.delay(Delay.kt:121)
	at kotlinx.coroutines.RunBlockingTest$dispatcherShutdownTest$j$1.invokeSuspend(RunBlockingTest.kt:15)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:33)
	at kotlinx.coroutines.DispatchedTask.run(DispatchedTask.kt:106)
	at kotlinx.coroutines.scheduling.CoroutineScheduler.runSafely(CoroutineScheduler.kt:571)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.executeTask(CoroutineScheduler.kt:750)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.runWorker(CoroutineScheduler.kt:678)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.run(CoroutineScheduler.kt:665)

[qwwdfsad]

I've improved the exception message and documentation.

The API is marked as delicate and has a pretty narrow use case, so I'm not sure it's worth (both in terms of contract and maintainability) giving more guarantees than now.