Add missed case for prefer_collection_literals

[natebosch]

Shows that the check for returning from a function literal is incorrect.

Description

Please include a summary of the change and a reference to any corresponding issues.
If this is a fix to an existing lint rule or a proposed new one, please include
the name of the rule in the pull request.

If this is a new lint or feature and there is not an existing tracking bug, please
consider opening one to better facilitate conversation.

Fixes # (issue)

[coveralls]

Coverage remained the same at 95.761% when pulling 24e499d on repro-collection-literal-false-negative-closure into a9f509a on main.

[srawlins]

I think someone is working on a "FAILED LINT" syntax for the linter test cases.

[bwilkerson]

It would be good to comment this line so that it's clear that the failure is incorrect.

Even better would be to write this as a reflective test so that it could be marked as a @failingTest with the correct expectations. (See https://github.com/dart-lang/linter/blob/main/test/rules/prefer_collection_literals_test.dart.)

Other than that, lgtm.

[bwilkerson]

I think someone is working on a "FAILED LINT" syntax for the linter test cases.

I hope not. We don't need it given that we've decided to switch over to using test_reflective_loader.

[srawlins]

I hope not. We don't need it given that we've decided to switch over to using test_reflective_loader.

Ah that's good, this PR can be converted to that, then?

[bwilkerson]

I wouldn't block it from being merged because it hadn't been converted, but that was my suggestion.

[srawlins]

I wouldn't block it from being merged because it hadn't been converted, but that was my suggestion.

But the introduced test is failing. How could we merge it?

[bwilkerson]

But the introduced test is failing. How could we merge it?

I didn't notice that. (I mean, who pays attention to the bots? :-) ) Then yes, it needs to be changed, so making it a reflective test makes the most sense.

[natebosch]

I'm pretty sure I got the offset correct...

[bwilkerson]

I suspect that the offset is wrong, but it still fails, so it's probably close enough. We have some support coming down the line that will make tests like this much easier to write.