Cleanup use of dicts, e.g. use of .keys()

[Flamefire]

Using x in dict.keys() is discouraged and should be x in dict directly which is potentially more performant and easier to read/shorter. Accesses to keys and their values should use dict.items to avoid the lookup overhead.

[jfgrimm]

could this not be simplified to the following?

Suggested change

	easyconfigs, files_to_delete, patch_files, py_files = [item[1] for item in sorted(paths.items())]
	easyconfigs, files_to_delete, patch_files, py_files = sorted(paths.values())

[Flamefire]

We need to sort by key so that we get the right order.
It looks like a dataclass or NamedTuple would be better than a dict where you know all existing keys.

[jfgrimm]

ah yes, of course

[Flamefire]

TBH: I had to read the code here and the caller again to understand this. Shall I add a commit changing this to a named Tuple? Even though it is a "public" method I'd doubt anyone would be using this outside of us.

[jfgrimm]

I would be fine with that

[Flamefire]

That bubbled up the callchain more than I expected: The origin of this dict is categorize_files_by_type
IMO still very much worth it: Makes it much easier to reason about what is actually passed instead of having a paths parameter in multiple places and sometimes it is a list of paths, and sometimes a dict with only some keys which might be misspelled.

If you'd like to review that change in isolation then use #5190 and I'll rebase this (will cause an easy conflict to resolve)