dispose Mocha Runner after use to avoid MaxlistenersExceededWarning

[boneskull]

This change eliminates the frequent MaxListenersExceededWarnings emitted when running gulp runtests-parallel. The Runner#dispose method removes all EE listeners the Runner instance created. When using the Runner manually like this, you'll want to call this function. It is a relatively new, and the Mocha typings have not been updated to reflect its presence (hence the @ts-ignore).

Apologies, I could not find an issue here associated with this change, but I'm sure there is one somewhere. I'll leave this as a draft until I find it. Created #41404.

Fixes #41404, mochajs/mocha#4329 reported by @weswigham

---

Commit summary:

• call runner.dispose() on listener for end event
• removed manual unhandledRejection listener as Mocha v8.2.0 now has one

[typescript-bot]

The PR doesn't have any linked issues. Please open an issue that references this PR. From there we can discuss and prioritise.

[DanielRosenwasser]

I've got a DefinitelyTyped PR over at DefinitelyTyped/DefinitelyTyped#49382 if you'd like to take a look so we can get rid of the // @ts-ignore comment.

[boneskull]

@DanielRosenwasser not sure if you're asking me to review the typings. if it works for you, it works for me... I haven't figured out how to consume them when working in Mocha itself

[boneskull]

but happy to update this PR when those changes land

[DanielRosenwasser]

@types/mocha@8.0.4 should have the right changes.

[DanielRosenwasser]

I've updated the PR, let's see if it works.

[DanielRosenwasser]

@amcasey are the build failures just a matter of updating the lockfile?

[amcasey]

@DanielRosenwasser I think I'm missing a step. Was dispose added recently? When I check out this branch, it builds locally, so I'm missing something if it's a package-lock.json issue.

Edit: I wasn't building the tests locally...

[amcasey]

Okay, I'm finally caught up. Yes, this PR branch is using @types/mocha@8.0.3, whereas master is on @types/mocha@8.0.4. I haven't seen many such issues, so I'd suggest we treat it as a one-off and just merge/rebase this PR to pick up the new types.

[DanielRosenwasser]

Thanks @boneskull!