Don't require a patch comment when the patch is a local file

[roberth]

We correctly ask for patches to come with a comment because we prefer to fetch them and we have no other place to provide context for the patch. See also https://github.com/jtojnar/nixpkgs-hammering/blob/master/explanations/missing-patch-comment.md

However, when the a patch is a local file, because it is not upstreamable, we have plenty of control and the file name or embedded message can cover all there is to know. See NixOS/nixpkgs#130102 (comment)

I would like to avoid a situation where we have nonsense commits and nonsense comments littering Nixpkgs, so I suggest we don't report missing comments when the patch is a path expression.

[jtojnar]

I do agree that in trivial cases like that a comment is unnecessary as it just duplicates the patch name.

However, there are cases where the that will not be enough. Example: https://github.com/NixOS/nixpkgs/blob/90cbe95037ff727540691436c7f8532ac602b376/pkgs/development/tools/build-managers/meson/default.nix#L19-L60

I prefer that a summary comments are in the Nix file so that the person updating the package can quickly guess if the file is still needed (after reading the release notes). The patch itself can contain more detailed description.

So I would prefer not to remove the check but rather reformulate the message such that it is clearer it is just a suggestion and viewer discretion is advised. For instance, we could add examples where comment would be pointless to the explanation.

Related: #27