dogfood: unicode-normalization takes 3 minutes to check

During the dogfood test ,clippy is run on itself to make sure its code has no clippy warnings.
During this process, clippy checks all the dependencies as well, however I think this may actually not be needed.

On my system, I can see that checking the unicode-normalization crate using clippy takes 2-3 minutes alone yet a simple "cargo check" of clippy seems to be much faster.

Perhaps we can somehow "cargo check" the dependencies in the dogfood test and only check the actual repo sourcecode using clippy for a decent dogfood speedup?

Start by locating the dogfood test that runs Clippy on the repository and inspect how dependencies are handled. Compare the current dependency checking with the reported cargo check behavior, using unicode-normalization as the timing case. Done means the dogfood test avoids unnecessarily running Clippy on dependencies while still checking the repository source.