8.2.0 does not work with sinon.useFakeTimers/setTimeout stubbing

[hldswrth]

Prerequisites

• Checked that your issue hasn't already been filed by cross-referencing issues with the faq label
• Checked next-gen ES issues and syntax problems by using the same environment and/or transpiler configuration without Mocha to ensure it isn't just a feature that actually isn't supported in the environment in question or a bug in your code.
• 'Smoke tested' the code to be tested by running it outside the real test suite to get a better sense of whether the problem is in the code under test, your usage of Mocha, or Mocha itself
• Ensured that there is no discrepancy between the locally and globally installed versions of Mocha. You can find them with: node node_modules/.bin/mocha --version(Local) and mocha --version(Global). We recommend that you not install Mocha globally.

Description

Mocha 8.2 introduces unexpected behaviour when using sinon.fakeTimers()

Steps to Reproduce

Tests which don't call done() are failing with a message saying that done() has been called multiple times. Other tests which are supposed to call done are ending before done() is called.

Here's an example:

const chai = require('chai')
chai.use(require('chai-as-promised'))
const expect = chai.expect
const sinon = require('sinon')
chai.use(require('sinon-chai'))
const AsyncLock = require('async-lock')
const P = require('bluebird')
const asyncLock = new AsyncLock({ Promise: P })

describe('timeout-test', () => {
  describe('using async-lock', () => {
    it('says done is called multiple times', () => {
      let clock = sinon.useFakeTimers()
      sinon.stub(clock, 'setTimeout').callsFake((delayedFunction, delay, ...args) => {
        delayedFunction(...args)
      })

      const rejectInLock = function () {
        return asyncLock.acquire('LOCK', () => {
          return P.reject(new Error('999'))
        })
      }

      return expect(rejectInLock()).to.be.rejectedWith('999')
    })
  })
})

Expected behavior: [What you expect to happen]
The test should pass. for 8.1.3 and below it works, 8.2.0 it breaks.

Actual behavior: [What actually happens]
The test fails with this message

Error: done() called multiple times in test <timeout-test using async-lock says done is called multiple times> of file ....\donetest.js
    at tryCatcher (server\node_modules\bluebird\js\release\util.js:16:23)
    at Promise._settlePromiseFromHandler (server\node_modules\bluebird\js\release\promise.js:547:31)
    at Promise._settlePromise (server\node_modules\bluebird\js\release\promise.js:604:18)
    at Promise._settlePromise0 (server\node_modules\bluebird\js\release\promise.js:649:10)
    at Promise._settlePromises (server\node_modules\bluebird\js\release\promise.js:729:18)
    at _drainQueueStep (server\node_modules\bluebird\js\release\async.js:93:12)
    at _drainQueue (server\node_modules\bluebird\js\release\async.js:86:9)
    at Async._drainQueues (server\node_modules\bluebird\js\release\async.js:102:5)
    at Immediate.Async.drainQueues [as _onImmediate] (server\node_modules\bluebird\js\release\async.js:15:14)
Error: 999

Reproduces how often: [What percentage of the time does it reproduce?]
100%

Versions

8.2.0

• The output of mocha --version and node node_modules/.bin/mocha --version:
  'mocha' is not recognized as an internal or external command, operable program or batch file.
  8.2.0
• The output of node --version:
  v10.22.0
• Your operating system
  • name and version: Windows 10
  • architecture (32 or 64-bit): 64 bit
• Your shell (e.g., bash, zsh, PowerShell, cmd): cmd
• Your browser and version (if running browser tests): N/A
• Any third-party Mocha-related modules (and their versions):
  "async-lock": "^1.2.0",
  "bluebird": "^3.4.0",
  "chai": "^4.1.2",
  "chai-as-promised": "^7.1.1",
  "mocha": "^8.0.1",
  "sinon": "^9.0.0",
  "sinon-chai": "^3.2.0",
• Any code transpiler (e.g., TypeScript, CoffeeScript, Babel) being used (and its version): None

Additional Information

[boneskull]

this may be fixed by #4489

[boneskull]

please try v8.2.1

[hldswrth]

I can confirm test passes with v8.2.1, so closing this issue, thanks.

I actually did like the uncaught error reports in v8.2.0, it forced us to check our code and tests very carefully for unhandled promise rejections and in a couple of cases did identify actual errors in our code.