Allow anonymous namespaces

[pkesseli]

Only give warnings for namespaces with actual names, in accordance with
ad41375.
This enables the use of anonymous namespaces without warnings.

[pkesseli]

Rebased @peterschrammel. Linter fails because of the regression tests for the linter.

[reuk]

How does this interact with namespace aliases i.e. namespace xyz = my::nested::namespaces;? From glancing at the code, it looks like aliases will be treated the same as declarations. The regression test should include some aliases to document that the current behaviour is desired. I don't think the coding standard mentions aliases, so either behaviour (allowing or disallowing) is valid, although the choice should be documented somehow.

[reuk]

Please document the expected linter behaviour in the presence of namespace aliases.

[pkesseli]

This pull request doesn't change the script's behaviour with respect to namespace aliases. They were marked as violations before and they are marked as violations with this new implementation too. I added an additional case in the regression test to confirm this.

If we want to explicitly handle or document namespace aliases, I thus suggest to do it in a separate pull request.

[reuk]

OK, makes sense. Thanks for adding the extra test case.

[tautschnig]

@pkesseli Would you mind adding "Fixes: #932" to the commit message so that the merge will auto-close issue #932?

[pkesseli]

Done.