Improve sc-service API

[nazar-pc]

This improves sc-service API by not requiring the whole &Configuration, using specific configuration options instead. RpcConfiguration was also extracted from Configuration to group all RPC options together.

We don't use Substrate's CLI and would rather not use Configuration either, but some key public functions require it even though they ignored most of the fields anyway.

RpcConfiguration is very helpful not just for consolidation of the fields, but also to finally make RPC optional for our use case, while Substrate still runs RPC server on localhost even if listening address is explicitly set to None, which is annoying (and I suspect there is a reason for it, so didn't want to change the default just yet).

While this is a breaking change, most developers will not notice it if they use higher-level APIs.

Fixes #2897

[nazar-pc]

Would ExecutorConfiguration be a good name for use with new_wasm_executor that combines the following options?:

• default_heap_pages
• wasm_method
• max_runtime_instances
• runtime_cache_size

This is probably the last thing that I'd like to stop using &Configuration for now.

[paritytech-cicd-pr]

The CI pipeline was cancelled due to failure one of the required jobs.
Job name: test-linux-stable 1/3
Logs: https://gitlab.parity.io/parity/mirrors/polkadot-sdk/-/jobs/7094356

[nazar-pc]

Resolved merge conflicts after #4792 and would really appreciate if someone takes a closer look here

[dmitry-markin]

The changes seem justified to me. Thanks again for streamlined per-commit review process!

[bkchr]

Two nitpicks.

[nazar-pc]

Applied suggestions

[niklasad1]

while Substrate still runs RPC server on localhost even if listening address is explicitly set to None, which is annoying (and I suspect there is a reason for it, so didn't want to change the default just yet).

I suppose you want to start your service without any rpc server then? Can you open an issue for it please?

I don't remember why we do it but that can changed if desired also with the latest refactoring it should be possible do addr: Some(vec![]) as workaround but I think it will throw an error when no rpc listen addr is found and bail early with the error.