=receptionist Fix receptionist crash on concurrent registrations; those are fine and expected

[ktoso]

New receptionist was too aggressive in what it deemed acceptable to be send onNext to the reception stream.

Concurrent messages are fine; specifically if [first:1] then [first:1, second:1] happens, these may have been concurrent in vector clock meaning, but it means that the event has more information so we should be emitting it

Actual fix is cc51dba (#1059) and the rest is supporting infra

resolves #1053

[ktoso]

Joining timeout in

19:32:33 Test Case 'ClusterSingletonPluginClusteredTests.test_remoteCallShouldFailAfterAllocationTimedOut' started at 2022-08-30 10:32:20.538
19:32:33 <EXPR>:0: error: ClusterSingletonPluginClusteredTests.test_remoteCallShouldFailAfterAllocationTimedOut : threw error "
19:32:33         try await self.assertMemberStatus(on: second, node: firstNode, is: .down, within: .seconds(10))
19:32:33                              ^~~~~~
19:32:33 error: MembershipError(awaitStatusTimedOut(10.0 seconds, Optional(MembershipError(statusRequirementNotMet(expected: DistributedActors.Cluster.MemberStatus.down, found: Member(sact://first:1722502985102675705@127.0.0.1:7111, status: joining, reachability: unreachable)), at: DistributedActors/ClusterControl.swift:252))), at: DistributedActors/ClusterControl.swift:338)"

[ktoso]

Seems to be same as #1034 - the nodes seem to sometimes need longer to join up for the test.

[ktoso]

Integration test is green which is good, this one would have triggered the bug this is fixing :-)

[yim-lee]

One typo but otherwise LGTM 👍

[yim-lee]

This reads very nicely 👍

[ktoso]

I recently realized we don't always have to do the __ trickery by the way :)

[yim-lee]

Ah, makes sense I think.