feat: rpc as a task with publishing signature

[ChaoticTempest]

• creates an RPC task where everything related to sending out requests to RPC services (whether NEAR or eth) are handled.
• moves over all the publishing of signatures to this RPC task
• publishing on either chains will be created as a separate task where each task will handle its own retrying as to not block the rpc task itself.
• new NearClient and EthClient for easier handling

Going to defer moving the other RPC related calls like join and reshare till later because those will require additional changes to the state machine for better handling of errors across a channel

[jakmeier]

The code changes mostly makes sense to me!

But can I just ask, what is the main goal? Is it mainly a refactor to clean up code architecture? Or is it trying to unblock certain tasks and introduce more parallelism? It's not quite clear to me from the PR description, it only describes what the code does, not why.

[ppca]

🚢 from perspective of eth respond() publishing.

[volovyks]

Good refactoring should make our lives easier when adding other chains. Other comments are reasonable.

[ChaoticTempest]

But can I just ask, what is the main goal? Is it mainly a refactor to clean up code architecture? Or is it trying to unblock certain tasks and introduce more parallelism? It's not quite clear to me from the PR description, it only describes what the code does, not why.

whoops, forgot to mention this in the PR description: it's mostly to unblock new signature from not progressing. Before this PR, signature manager would try to make an RPC call and block the whole entire main loop from progressing. This was especially bad in the case that an error occurs during publishing and we would exponentially retry to publish while still in the signature_manager context, which meant other tasks couldn't progress or more signatures could not be processed.

[volovyks]

Seams useful to me.

[ChaoticTempest]

it's logged elsewhere