minimize setup for BesuCommandTest

[macfarla]

• keyPair → static + moved initialization to a new BeforeAll static initKeyPair() — EC key derivation from a fixed private key now happens once per test class instead of 188 times
• Mock(lenient = true, answer = Answers.RETURNS_SELF) on mockRunnerBuilder and mockControllerBuilder — all fluent builder chain methods now self-stub automatically
• Removed ~60 when() stubs from initMocks() — only the non-self-returning stubs remain (build(), fromEthNetworkConfig(), controller/protocol context getters)
• Removed duplicate GlobalOpenTelemetry.resetForTest() from inside parseCommand() (already done in BeforeEach setUpStreams())

Is there any risk of tests not being independent with these changes?

keyPair → static: No risk. It's read-only after initialization — no test ever mutates it. It was always derived from the same fixed private key hex, so making it static just removes redundant recomputation of the same value.

RETURNS_SELF on the builder mocks: No risk. MockitoExtension re-creates all @Mock fields from scratch before every test - RETURNS_SELF is just the default answer applied to the fresh instance. Each test still gets a clean, independent mock.

mockControllerBuilder.build() explicit stub: Safe. Explicit when().thenReturn() stubs always take precedence over the default answer in Mockito. RETURNS_SELF only applies to methods that aren't explicitly stubbed.

One subtle thing worth knowing: RETURNS_SELF returns the mock for any method whose return type is assignable from the mock's type. This means if a test were trying to verify that a builder method was called with specific arguments and checking the return value behaves like null/empty, it could be masked. But:

• All tests that verify(mockRunnerBuilder).someMethod(...) still work correctly — RETURNS_SELF doesn't affect verification
• The previous stubs used any() matchers anyway, so they already intercepted everything regardless of arguments

The one pre-existing independence issue we didn't change: rpcEndpointServiceImpl is a static final instance shared across all tests. If a test registers a custom RPC endpoint on it and doesn't clean up, it leaks to subsequent tests. But that was already the case before our changes.

./gradlew :app:test --tests "*BesuCommandTest*"
local run -
before: 5min
after: 4min

[Copilot]

Pull request overview

This PR reduces per-test setup overhead in BesuCommandTest by caching deterministic crypto setup and simplifying Mockito fluent-builder stubbing.

Changes:

• Move deterministic KeyPair derivation to a single per-class initialization (@BeforeAll) and make it static.
• Configure fluent builder mocks to use Answers.RETURNS_SELF, removing many redundant when(...).thenReturn(this) stubs.
• Remove redundant GlobalOpenTelemetry.resetForTest() call from parseCommand().

---

You can also share your feedback on Copilot code review. Take the survey.

[usmansaleem]

LGTM. copilot suggestion regrding static final Keypair also make sense. You can further avoid @BeforeAll.