test: restore unstable network tests

[0x009922]

Context

Close #5153

Solution

Implement a TCP relay as described in #5164 (comment). This allows to suspend packages being sent to and from specific peers.

Checklist

• I've read CONTRIBUTING.md.
• (optional) I've written unit tests for the code changes.
• All review comments have been resolved.
• All CI checks pass.

[0x009922]

Relay somewhat works, but weird network behaviour continues happening. Specifically:

• "Suspended" peers still find way to broadcast transactions?
• Network synchronizes for a long time for no clear reason
• Flaky. Sync sometimes doesn't happen. Not sure if something is wrong with the relay or with Iroha itself

[mversic]

Relay somewhat works, but weird network behaviour continues happening. Specifically:

• "Suspended" peers still find way to broadcast transactions?
• Network synchronizes for a long time for no clear reason
• Flaky. Sync sometimes doesn't happen. Not sure if something is wrong with the relay or with Iroha itself

I think it's good enough to merge. Then we'll have to backtrack and compare to last known working iroha build

[0x009922]

I think it's good enough to merge. Then we'll have to backtrack and compare to last known working iroha build

I have some doubts that there might be something wrong with the relay.

Anyway, let me rehearse the code and I will undraft it.

[s8sato]

What do you think about renaming "faulty peers" to "fault tolerance", which sounds to me like a clearer expression of what this test should guarantee

[0x009922]

Sounds good to me.

[0x009922]

Status of this PR

The introduced tests are flaky, and something not clear is going on with the tested networks. It requires a deeper investigation.

I am personally now focusing on updating hyperledger-iroha/iroha-javascript, and postponed work on this PR until better times.

The relay is probably working correctly. However, I wouldn't merge this PR as it is now, as it will make CI flaky. Maybe it would make sense to merge this PR with the flaky tests being ignored, but I don't see a good reason to make this semi-step.