feat(rule): disallow unnecessary async function wrapper for single await call

[hainenber]

Closes #1722

I've implemented said rule to lint for unnecessary async wrapper for single await call inspired (or blatantly copied? :D) from @G-Rath's ESQuery selector in linked issue.

This is also put into recommended option as suggested. PTAL when you have time, folks!

[G-Rath]

We can't change our configs without doing a new major, so please remove this from recommended

(On mobile so can't currently do a full review, but figured I'd this in in case you get to it before me 🙂)

[hainenber]

I've done removing this rule from recommended config. Aside from a crashed Jest worker in MacOS test setup, this is ready for review :D

[G-Rath]

could you clean up the indentation of the code samples? the code should be indented one level from the indent level of the dedent, and the end backtick should be at the same indent as the dedent

i.e.

Suggested change

	dedent`
	it('pass', async () => {
	expect();
	})
	`,
	dedent`
	dedent`
	it('pass', async () => {
	expect();
	})
	`,
	dedent`

[hainenber]

Understood. I'll amend in next commit.

[G-Rath]

we've got a utility for this

Suggested change

	(awaitNode?.type !== AST_NODE_TYPES.ArrowFunctionExpression &&
	awaitNode?.type !== AST_NODE_TYPES.FunctionExpression) ||
	!awaitNode ||
	!isFunction(awaitNode) ||

[hainenber]

TIL. I'll amend in next commit

[G-Rath]

What do you think of a phrasing more like "Unneeded async function"? that way we don't have to deal with the "rejected/resolved", and I think it's a bit more positive to be conveying we think you can have less code as an optimization

[hainenber]

I'm cool with that 👍

[hainenber]

All amended in newest commit :D

[G-Rath]

I think this is pretty much there, just got a few more tests I'd like - also, I'm wondering if a name like no-unneeded-async-expect-function might be a better name?

I think the use of "unneeded" is good because it indicates the rule is about an optimization that lets you have less code, rather than guarding against a particular bug or gotcha

[G-Rath]

I think it would be better to say something like "This rule triggers a warning if expect is passed an async function that has a single await call", since that is what "unnecessary" means in this context

[G-Rath]

it would be good to have some tests that:

• pass a non-async function to expect
• pass a reference to another function which is async
• uses await within the expect call (e.g. expect(await x()))
• uses of other matchers, to show that we don't actually care if rejects or resolve is being used
• use of await within an array

[G-Rath]

nit: these would be more correct as "should be fixed" 😅

[G-Rath]

it'd be good if we could have a test with say a Promise.all, to help prove the fixer works in more complex situations