Failure to discover default value for a primitive should not lead to document its default value

[snicoll]

See #46427 for a concrete example.

If we have a primitive with a default value we can't infer, then its default is documented. We should rather not document the default at all.

[Kguswo]

#46551

I just tried to fix the problem. Any feedback would be appreciated.

[snicoll]

Closing in favor of PR #46551 - Thanks @Kguswo