Finish block building process even when cancel request is found

[ferranbt]

📝 Summary

This PR introduces three changes:

• It refactors some of the common utilities shared among the tests into a single TestHarness struct which spins the test framework with all the components. Note that it does not update the tests to use this new utility.
• It fixes an issue with the block builder that would stop the block building process and not return any block if a cancel request was found. This happens when an FCU and a getPayload request are called to close to each other, the getPayload cancels the block building process, and getPayload waits forever for a block that will never be built. Now, the block building finishes.
• It adds an integration test to cover this use case with the new utility.

💡 Motivation and Context

---

✅ I have completed the following steps:

• Run make lint
• Run make test
• Added tests (if applicable)

[Copilot]

Pull Request Overview

This PR addresses a bug in the block building process where a cancel request prevented the process from completing by refactoring the block builder logic. Key changes include:

• Introducing a new constructor (new_with_port) and updating function signatures to pass EngineApi by value rather than by reference.
• Removing the early cancellation check in the payload builder so that block building always finishes.
• Adding a new integration test and a TestHarness utility to better simulate and validate the block building process.

Reviewed Changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
crates/op-rbuilder/src/tester/mod.rs	Added new_with_port and updated BlockGenerator to use value types; added logging and refactored submit_payload with an extra no_sleep parameter.
crates/op-rbuilder/src/tester/main.rs	Updated BlockGenerator instantiation to pass EngineApi by value.
crates/op-rbuilder/src/payload_builder_vanilla.rs	Removed the cancelled check from execute_best_transactions call.
crates/op-rbuilder/src/integration/mod.rs	Introduced TestHarness and port allocation helper along with related refactoring.
crates/op-rbuilder/src/integration/integration_test.rs	Added an integration test for the edge case with close FCU/getPayload calls.
crates/op-rbuilder/Cargo.toml	Added parking_lot dependency.

Comments suppressed due to low confidence (2)

crates/op-rbuilder/src/tester/mod.rs:344

• [nitpick] Consider refactoring the submit_payload function to reduce the parameter count; bundling the delay-related parameters into a configuration struct could improve readability and maintainability.

no_sleep: bool, // TODO: Change this, too many parameters we can tweak here to put as a function arguments

crates/op-rbuilder/src/integration/integration_test.rs:363

• Consider adding a timeout mechanism for the submit_payload call in this integration test to ensure that the test does not hang indefinitely in case of an underlying issue.

// TODO: In the fail case scenario, this hangs forever, but it should return an error

[avalonche]

the getPayload cancels the block building process, and getPayload waits forever for a block that will never be built

this would only happen if fcu causes a cancel before getPayload is resolved right? why would getPayload cause a cancel?