doc: Add docs for EthApi

[tcoratger]

Should close #9540.

[mattsse]

lgtm, pending @emhane

[emhane]

this is too much chagpt fluff

[tcoratger]

Hey @emhane,

Just to tell you that in fact I used ChatGPT (that seems normal to me for generating documentation in 2024, right?). However, I spent about 30 minutes writing the entire OpEthApi example (in documentation form) with also the definition for myself of each helper trait:

/// - `EthApiSpec`: Defines basic Ethereum network information and status.
/// - `SpawnBlocking`: Handles task spawning and concurrency.
/// - `LoadFee`: Manages fee-related data loading and caching.
/// - `Call`: Handles EVM call operations.
/// - `LoadState`: Manages state loading and caching.
/// - `EthState`: Provides Ethereum state-related functionality.
/// - `EthCall`: Implements Ethereum call-related methods.
/// - `EthFees`: Handles fee estimation and history.
/// - `Trace`: Provides tracing functionality for transactions.
/// - `AddDevSigners`: Manages development signers for testing.
/// - `UpdateRawTxForwarder`: Allows updating of raw transaction forwarding behavior.

that ChatGPT cannot guess (at least not from what I understood). Then in fact when I pass all this context to ChatGPT (i.e. more or less all the main documentation of the EthApi), he is much better than me at organizing the pieces and writing clear documentation (and I imagine that no matter who by the way).

But in general, I don't really understand why you close PRs without review and feedback loops. Every time I feel like I'm doing a lot of work for nothing. This is not a criticism at all, if you feel that this PR should not be merged, you are the one responsible for the code, I can only agree but I would like to understand better:

• On this PR we could have removed the parts that don't suit you, shortened it, removed information, added what seemed missing to you.
• Here refactor: use BundleBuilder as a state builder #8861 it's the same I didn't really understand because there was no ChatGPT and a lot of super useful work to do on it. I saw a comment saying that we probably had to modify the BundleBuilder API but that's not a problem, we could have done it. But ultimately it was abandoned and everything was closed?

cc @mattsse