concurrent-api: Add ability to cleanup resources on timeout

[bryce-anderson]

Motivation:

The Single.timeout(..) family of operations are part of a class of operations will short-circuit the response based on the cancellation pathway, specifically they will return a TimeoutException down the error path. Because cancellation and the result are intrinsically racy there is currently the possibility that the result will be generated but we cannot return it to the subscriber having already given them an Exception in the error pathway. For values that are stateful this can result in a resource leak.

Modifications:

• Add a timeout combinator that includes a resource cleanup hook.
• Use that hook in the AbstractTimeoutHttpFilter where we need to drain the message body which can include live and stateful resources such as cleanup hooks and even the life cycle management of a HTTP session.

Result:

One less leak and at least the option for users to cleanup stateful resources on timeout races.

[bryce-anderson]

Note that this is one of two leaks discovered in the CapacityLimiter lock-up investigation. More details can be found in #3033.

[bryce-anderson]

Offline discussion: this will fix the problem but opens up the can of worms that many of our operators are not resource safe so it doesn't make sense to only address it here. We'll find another way.