Add a test to detect duplicated lint rules

[Atry]

HHClientDuplicatedLintErrorTest ensures that no lint error is found by HHClientLinter in examples for SingleRuleLinters, closing #409.

[fredemmott]

It's fine to have two errors for the same node, just as long as they're not talking about the same problem - e.g. await-in-a-loop with an always-false condition is a valid combination.

I think you're already handling this with the HHAST_IGNORE_ALL in the tests, but just wanted to make sure?

If so, it would be good to add an explanation comment to the test case itself about this, and saying use HHAST_IGNORE_ALL in tests to suppress false positives

[fredemmott]

Are we raising 0, or does this have a magic meaning?

[Atry]

It means this file does not type check.

[fredemmott]

Why does it have this meaning? hh_client doesn't pay attention to it, and it's not really intuitive.

Does this have special meaning to HHClientLinter, affecting non-lint errors? If so:

• this behavior should probably be removed
• HHClientLinter should probably always ignore any errors out of the 5xxx range:
  • all linters are in the 5xxx range
  • 1xxx-4xxx should be handled via /* HH_FIXME */ + hh_client - they shouldn't affect hhast at all (beyond whether or not hhast can parse the file)

[Atry]

It means a bug in the OCaml linter. facebook/hhvm#8950

[fredemmott]

IMO we should not report this as a lint error, and we should not require suppression - similarly to how we skip files that hh_parse can't deal with - unless the specific file is requested.

[Atry]

It's fine to have two errors for the same node, just as long as they're not talking about the same problem - e.g. await-in-a-loop with an always-false condition is a valid combination.

I think you're already handling this with the HHAST_IGNORE_ALL in the tests, but just wanted to make sure?

If so, it would be good to add an explanation comment to the test case itself about this, and saying use HHAST_IGNORE_ALL in tests to suppress false positives

Comments added in b94f654