Fix regression where function overload candidates might not be reported

[marsupial]

Description

597b6b1 introduced a bug which might make the error message less verbose than it used to be.
This rectifies that.

Tests

oslc-err-funcoverload

Checklist:

• I have read the contribution guidelines.
• I have previously submitted a Contributor License Agreement.
• I have updated the documentation, if applicable.
• I have ensured that the change is tested somewhere in the testsuite (adding new test cases if necessary).
• My code follows the prevailing code style of this project.

[lgritz]

Something is wrong, it's failing the CI tests, no longer correctly compiling some of the MaterialX shaders.

[marsupial]

Yes, was not using the proper constant.
Running on a separate CI before updating the PR.

[marsupial]

Should be good to go. Was actually trying to fix some other issues that are coming to light, but just reverted to simple fix for missing labels of some ambiguities.

[marsupial]

Might be prudent to revert #837 (597b6b1).
Not sure how it got missed, but I'm seeing some behavior that is definitely incorrect.

[lgritz]

So, just to clarify:

You think we should revert #837, but keep this one #843?

[lgritz]

Or did you mean we should put both on hold until you sort it out?

[marsupial]

I was thinking reverting #837 and rolling this (#843) into either a new PR or re-open #837.
If you're more interested in not reverting and moving forward with fixing the original problem I'll have a few more hours today and tomorrow (which should be enough to nail it down), but a back and forth might be necessary.

I'll do a further writeup of the problems I'm discovering in the next hour.

[lgritz]

OK, since we know there's something wrong, I'll revert #837 in master, and you can close this PR, and then when you have it totally worked out, just post a new combined PR (or re-open 837). There's no rush on this, we've obviously lived with the existing ambiguous edge cases for a long time with nobody complaining.