Aaron Stannard

Looks to me like this change definitely doesn't quite cut the mustard for some of the racy specs here - I'll need to harden the implementation some.

# RemotePingPong ## `dev` OSVersion: Microsoft Windows NT 10.0.19041.0 ProcessorCount: 16 ClockSpeed: 0 MHZ Actor Count: 32 Messages sent/received per client: 200000 (2e5) Is Server GC: True Thread count: 112...

No test failures, but several parts of the test suite timed out. **edit**: makes me wonder if there's an "unclean shutdown" problem with the new scheduler as-is.

actually, seeing quite a few failures in the Akka.Streams testsuite on my local machine

I'll give this a go - otherwise I'll give wait handles a try.

Let me take a look at this...

Worth noting that this has been rewritten on the `dev` branch to use `Task` and `async` rather than `Thread.Sleep` - cc @Arkatufus

testing to see if there are any other issues we need to fix in https://github.com/akkadotnet/akka.net/pull/5861 before we make this the default.

Now that #5861 has been merged into `dev` this should be a much more interesting test

https://github.com/akkadotnet/akka.net/issues/5385#issuecomment-1119566953 - ran some benchmarks for this change. Looks like a ~50% drop in Akka.Remote throughput under this configuration... That's a bitter pill to swallow.