feat: [UTP-242] [v2] Make subnet replica version of latest block available to execution environment

[michael-weigelt]

Alternative to #2082

This PR makes the replica_version part of RegistryExecutionSettings so that the execution environment has access to it when processing messages.

This PR will have a follow-up to make replica_version available via a management canister call as specified here.

[crusso]

Approving superficial changes to drun (do you agree @luc-blaeser?)

[adambratschikaye]

This is a bit confusing to me. Why are we putting it in the RegistryExecutionSettings if it doesn't come from the registry? Maybe instead it would make more sense as a new argument to execute_round? Or another option would be to combine the other arguments to execute_round that have info on the specific block into a single struct argument.

Then I think you also wouldn't need to pass the ReplicaVersion through all the message routing logic to read the registry.

[adambratschikaye]

Ah I see now where this is coming from after looking at the old PR. I'd say do either:

1. Add it as a new argument to execute_round and execute_subnet_message.
2. Combine it with idkg_subnet_public_keys into a struct which is passed to both execute_round and execute_subnet_message.

[michael-weigelt]

I share your concerns. The previous PR writes batch.replica_version it into the replicated state (metadata) and uses it from there, but it was questioned whether that is necessary. I believe RegistryExecutionSettings was suggested because it's already being passed along.

But I can instead change the signatures of all methods along the way and introduce a new, conceptually clean argument for this data.

[michael-weigelt]

Here is another draft, PTAL:
#2248