Elastic scaling: zombienet tests

[sandreim]

The following cases need to be covered:

• 2s blocks with MVP still works with new validator and cumulus nodes
• 2s RFC103 enabled parachains work with new validators and cumulus nodes
• 2s blocks with old + new cumulus honest nodes
• Negative test for collators sending it's own collation to all backing groups - candidates are rejected and parachain still manages to get 2s block times.

[alindima]

2s blocks with MVP still works with new validator and cumulus nodes
2s RFC103 enabled parachains work with new validators and cumulus nodes

These are part of #6452.

2s blocks with old + new cumulus honest nodes

I tested this locally. As expected, this doesn't achieve full throughput because the two different collators end up building on the same core at times. However, the chain doesn't stall, which is good. I'll see if I can add this to #6452 without it being too flaky.

Negative test for collators sending it's own collation to all backing groups - candidates are rejected and parachain still manages to get 2s block times.

For this we need a malus collator that will submit its collation to multiple backing groups. We should probably do this in a generic way because there are other scenarios that we want to test also #4526, and especially needed for the collator protocol revamp

[alindima]

I tested this locally. As expected, this doesn't achieve full throughput because the two different collators end up building on the same core at times. However, the chain doesn't stall, which is good. I'll see if I can add this to #6452 without it being too flaky.

Decided to remove this test, since we'll make these v1 candidates that send UMP signals invalid (in #6939)