AATH RFC0211 tests failing when running sequentially

[genaris]

In last AATH runs there are several tests failing related to #1391, but I noticed something weird in RFC 0211 tests: a test is now failing due to a timeout where it didn't previously.

When running locally, test passes if it's run individually. But if I run the whole suite or all RFC0211 (e.g. executing ./manage run -a javascript -d javascript -t @RFC0211 -t @RFC0160) , it gives this timeout when attempting to restart the agent.

Initially I thought that it was due to a regression introduced in #1413, but after inspecting the code I did not find any breaking change for this particular test. Later I found out that if I downgrade ws to 7.4.6 (the version used before #1388, which was merged the same day), the tests pass.

So probably it's related to a breaking change in ws module. In 8.0.0 release notes there are a few that can be good candidates for this issue: for instance, closing the WebSocketServer does not close existing inbound connections, which should be explicitly terminated by calling ws.terminate(). I did apply this particular change without success in this case, although I think it should be fixed in WsInboundTransport.close().

[TimoGlastra]

Interesting. Maybe we can downgrade ws for now, and then at a later stage updgrade it and add a test case that reproduces it in AFJ, so we can then fix it?