test: clarify qbft_manager harness setup

[diegomrsantos]

Problem, Evidence, and Context (Required)

• The qbft_manager tests currently hide simple scenarios behind generic setup and generic harness entrypoints, which makes behavior-focused reviews harder than they need to be.
• This is worth doing now because follow-up QBFT lifecycle and retention work is easier to review once the harness reads in terms of duties, clusters, and expected outcomes.
• Evidence: a reader has to jump through setup_test(...), raw TestContext::new(...), and verify_consensus() to understand straightforward scenarios like "start one instance and assert consensus".
• Relevant links: related review context in refactor: use beacon chain deadlines for QBFT instance cleanup #719.

Change Overview (Required)

• Rework the test helper surface so the high-level qbft_manager tests use intent-named helpers for runtime setup, beacon-vote cluster startup, and consensus assertions.
• Keep the existing harness behavior intact by preserving the executor lifetime semantics inside the test context instead of relying on an opaque setup bag.
• Update the existing manager tests to read as scenario-driven Arrange/Act/Assert flows instead of harness plumbing.
• Intentionally did not change any runtime QBFT behavior, protocol semantics, or production code paths.

Risks, Trade-offs, and Mitigations (Required)

• Main risk: accidentally changing test-harness behavior while simplifying names and setup flow.
• Trade-off: the test module grows a few thin wrapper helpers in exchange for much clearer behavior-level tests.
• Mitigation: the diff stays in a single test file, the underlying harness remains the same, and the full qbft_manager test crate still passes after the refactor.

Validation (Required)

• cargo test -p qbft_manager
• cargo fmt --all --check
• Commit hook checks: cargo fmt --all, cargo clippy --all, cargo sort workspace

Rollback (Required for behavior or runtime changes; optional otherwise)

• Safe to revert as a normal code-only rollback.
• No config, data, or operational impact.

Blockers / Dependencies (Optional)

• No functional blockers.
• This branch is currently 4 commits behind unstable; rebasing before merge should be straightforward because the PR only touches one test file.

Additional Info / Next Steps (Optional)

• This is intended as a prep refactor so follow-up qbft_manager test and lifecycle changes can build on clearer test intent.