SRI most important, but NOT critical bugs

[GitGab19]

During last call we agreed about focusing the next release on Integration Test Framework development and protocols documentation and refactoring. We fixed a lot of crucial bugs during last two release (1.0.1 and 1.0.2), but there are still some bugs already logged which can be defined important, even if they don't require any urgent fix.

This issue serves as an umbrella for them.

• Custom transactions from JDC removed from mempool during update #840
  - Since we mined many blocks during last testing, without getting this issue, I'm confident it doesn't require an urgent fix
  - It started to be tackled, but we definitely got too many back-and-forth here
  - To properly test a fix, it would need a proper integration test, so I would wait for it to be added to the new integration tests framework (it means I wouldn't include this issue in the next release)

• TProxy responding to mining.submit with result:true for low difficulty shares. #1006
  - This is not urgent, but it's an important one.
  - It's pretty straightforward, I can take it if we want to include it in the next release

• JDC does not receive shares anymore from Translator if two new blocks are found in few seconds #920
  - This has already been tackled with a temporary fix
  - For this reason it's not an urgent thing to put in the next release

• Validate incoming MiningJob #986
  - This is important, but I would not define it a bug, it's a feature that we always missed to add until now, and I don't see the reason why we should prioritize it now
  - It's been assigned to @Shourya742, but if he didn't started to work on this, I would say we could skip it for now. Otherwise we can include it in the next milestone

[rrybarczyk]

When introducing integration tests, we need to be very careful to keep everything in scope and prevent snowballing into another refactor fork. We can start on this by keeping the roles as close as possible to their current existence. Knowing that after the primary refactor of the protocols (#845) and then eventually the roles refactor will have downstream effects on the integration tests that are developed before this effort is finished. We should be careful how deep we go here in this meantime.

[rrybarczyk]

Eventually, we should have proper integration tests to completely replace the message generator for testing. Message generator will not be completely deprecated as it is still useful. Perhaps someone in the community wants to run Sv2 with go for example, then message generator is still useful for testing the compatibility for SRI. But as far as SRI development goes, it is far better to test inside Rust. All sanity checks should be done inside Rust.