Fix rpaths patcher being run on symbolic links

[mbargull]

Description

Use skip_symlinks=True for former is_codefile/codefile_type calls.
This just mechanically adds skip_symlinks=True to the calls sides that previously had this implicitly (before 1dc2470).

Fixes gh-5179 .

Checklist - did you ...

• Add a file to the news directory (using the template) for the next release's release notes?
• Add / update necessary tests?

[mbargull]

This currently targets 3.28.x since 24.1 is not yet used in conda-forge.
We should do one of these:

1. Create new patch releases for 3.28/24.1,
2. Create a patch release for 24.1 and use build/use that package on conda-forge.
3. Add https://github.com/conda/conda-build/pull/5181.patch as a patch in conda-forge/conda-build-feedstock and build a 3.28.4.post1 in case we can't do 3.28 release and can't use 24.1 in production yet.

This bug is a bit problematic since it doesn't cause builds to outright fail, but rather leads to silently create partially broken packages..

[kenodegard]

wow this was sneaky, I went back through the old PRs and agree that these are the old is_codefile/codefile_type calls

I'm in favor of doing another 24.1 patch, I'd prefer not to do another 3.28 unless we absolutely have to

[mbargull]

wow this was sneaky

Indeed -- kind of fortunate that we ran into the patchelf bug to notice it :).

I'm in favor of doing another 24.1 patch, I'd prefer not to do another 3.28 unless we absolutely have to

How about we merge this, release 24.1.2 and build it on conda-forge today and only if something else breaks, we look at 3.28 again?