nix: make usage in non-flake setups easier

[Sorixelle]

using the nix expressions in setups where flakes aren't in use was possible, but a little unwieldy (you had to reconstruct the overlay defined in the flake to get copyparty in your nixpkgs set). this pr moves the overlay into it's own dedicated file, allowing it to be used directly without requiring something like flake-compat. there's also some extra documentation, illustrating how to use the module in a non-flake nixos setup.

i've also added a commit that removes a (seemingly) unnecessary evaluation of nixpkgs, to avoid contributing to the "1000 instances of nixpkgs" issue. this one might be a little controversial, so i've added it as a suggestion - if you're not interested in this change, i'll happily drop the commit. it's not super necessary for anything, just a potential optimisation i spotted.

this pr complies with the dco; https://developercertificate.org/

[chinponya]

LGTM. I don't see a reason why it should be dropped.

It might just conflict a little with #253 depending on which one gets merged first.

[Sorixelle]

the main issue with that appliedOverlay trick is that it doesn't work when the overlay expects final to be the actual final nixpkgs set (ie. it uses final to reference things the overlay itself defines, or things from other required overlays). that's not happening here though, so i figured it would be fine, but if the overlay does get a little more complex and relies on final actually being what it should, you'd probably have to drop that trick and just eat the extra instantiation.