🐛 Bug: `--fail-hook-affected-tests` drops nested-suite tests when a `beforeEach` hook fails

[spokodev]

Bug Report Checklist

• This is NOT a security, npm audit, or GitHub Advisory issue.
• I have read and agree to Mocha's Code of Conduct and Contributing Guidelines.
• I have searched for related issues and issues with the faq label, but none matched.
• I have 'smoke tested' the code to be tested by running it outside the real test suite.
• I want to provide a PR to resolve this.

Expected behavior

--fail-hook-affected-tests (added in v12) is documented to report tests as failed when they are affected by a hook failure, instead of silently skipping them. A failing beforeEach affects every test the hook would have run before, including tests in nested describe blocks, so all of them should be reported as failed. This is what before/beforeAll already does.

Actual behavior

With --fail-hook-affected-tests on, a failing beforeEach only fails the remaining tests of the immediate suite. Tests inside nested sub-suites vanish from the run entirely (not failed, not pending), so the failure and total counts are under-reported in every reporter (spec, json, tap, xunit).

Minimal, Complete, and Verifiable Example

// repro.mjs
describe('outer', function () {
  beforeEach(function () { throw new Error('boom'); });
  it('direct test', function () {});
  describe('nested', function () {
    it('nested test 1', function () {});
    it('nested test 2', function () {});
  });
});

Run: mocha --fail-hook-affected-tests repro.mjs

beforeEach result: 2 failing (the hook and direct test); the two nested tests are dropped.

Swap beforeEach for before: 4 failing (the hook, direct test, nested test 1, nested test 2) are all reported. That is the behavior the beforeEach path should match.

Versions

• mocha 12.0.0-rc.1 (also reproduces on current main)
• Node.js 24
• Platform-independent (reproduced on macOS)

Additional Information

before/beforeAll already recurses into sub-suites via its failAffectedTests path; the beforeEach branch fails only the immediate suite's not-yet-run tests and never recurses. The two branches should be consistent.