Fix misnamed parameter in stubtest_third_party

[Avasam]

I spotted this a while ago, but then forgot about it again.
Seems like when I introduced this in #9173 we totally glanced over that the name of the parameter is completely wrong, it's not specified "stubs", but specified "platforms" (in METADATA.toml)

[AlexWaygood]

It's hard to find a great name for this flag. I interpreted --specified-stubs-only as being shorthand for "only run stubtest on the stubs that specify they're tested on this platform in their METADATA.toml files"

[Avasam]

It's hard to find a great name for this flag. I interpreted --specified-stubs-only as being shorthand for "only run stubtest on the stubs that specify they're tested on this platform in their METADATA.toml files"

I totally meant it as "only run if the current platform is specified" with a focus on the platform.

The problem with the name --specified-stubs-only is that it implies only specified stubs are ran. Which stubtest already does by virtue of its dists arguments.
Your interpretation makes sense, but the problem I see is that it's not clearly implied what's specified: "only run stubtest on the stubs that specify they're tested on this platform in their METADATA.toml files", vs. "only run stubtest on the stubs that specify they're tested (let's say) without ignoring missing stubs in their METADATA.toml files" for example.

A more realistic example of specifying something that reuses the same flag would be adding a python version specification. But I'd rather renamed it specified-system-only then.

At the end of the day it's a flag meant only to be used by the CI, not directly by devs. And there's a description that comes with it. So it's not major. But it was a typo from what I originally meant (the kind of typo where you think about something else and end up subconsciously typing that word instead). So it can either be fixed, or retroactively change the meaning/semantic I meant and it's no longer a typo 😉

[JelleZijlstra]

The new name does seem better from a cursory reading of the code.