Fix potential recursion between users.users and home-manager.users

[Bert-Proesmans]

Description

Changes the strategy how home-manager integrates with users.users.<name>.packages to pull instead of push. This side-steps the potential circular dependency when the attribute sets home-manager.users or users.users are calculated from each other.

Checklist

• Change is backwards compatible. (AFAIK, please doublecheck edge cases)

• Code formatted with ./format.

• Code tested through nix-shell --pure tests -A run.all
  or nix build --reference-lock-file flake.lock ./tests#test-all using Flakes.

• Test cases updated/added. See example.

• Commit messages are formatted like

  {component}: {description}
  
  {long description}
  

  See CONTRIBUTING for more information and recent commit messages for examples.

• If this PR adds a new module

  • Added myself as module maintainer. See example.

Maintainer CC

[khaneliman]

error: attribute 'enable' missing
       at /home/runner/work/home-manager/home-manager/nixos/common.nix:56:10:
           55|       config = mkIf
           56|         (config.enable && cfg.useUserPackages && (lib.hasAttr name cfg.users)) {
             |          ^
           57|           packages = [ cfg.users.${name}.home.path ];

[Bert-Proesmans]

I based that if-test on the users options on unstable. looks like 24.11 does not have the "enable" option backported.
supposing we want backwards compatibility accross nixpkgs versions, is the preference to remove that check or do i also validate if 'enable' is even an option?

[Bert-Proesmans]

I checked the impact of setting on users.<name>.packages and it doesn't look like anything could go wrong, so removing the check for backwards compatibility should be ok.
Now, this change manipulates a non-owned option set so it might be a good idea to look into future proofing. Home-manager shouldn't generally do anything to the user if that user is disabled AFAIK, and "enable" is a standardised/intuitive option (so won't be changed in the future). This is an argument to keep the check to prevent future (semi-)related breaking and prevent future unwanted side-effects.

I rebased the commits per checklist instructions. The tests do not all succeed, can't directly say if this error is related to this change (doesn't look so at first sight);

error: attribute 'applicationName' missing
       at /nix/store/g4ppspdl4fy7hnp4jgjl4ll03v7i08w3-source/pkgs/applications/networking/browsers/firefox/wrapper.nix:167:23:
          166|         inherit icon;
          167|         desktopName = browser.applicationName;
             |                       ^
          168|         startupNotify = true;

EDIT; Regarding tests, I get failures different from CI on my local machine. Is there impurity in the testset? I'll need some guidance on how to tackle the errors.