azure-cli: install completions for all shells

[maxbrunet]

This switches to use register-python-argcomplete directly to generate and install the completion similarly to other Python tools like pipx.

The az.completion.sh script is not the actual completion, but runs eval: https://github.com/Azure/azure-cli/blob/azure-cli-2.60.0/src/azure-cli/az.completion.sh#L11

The previous state did not autoload for zsh and missed the installation for fish.

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[SFrijters]

For me this breaks cross compilation nix build .#pkgsCross.aarch64-multiplatform.azure-cli -L in new and interesting ways: generating the autocompletions recursively spawns default-builder.sh processes until OOM. This is after applying a workaround described in #338231 to even get to that point so I'm not sure if this is important at the moment. Maybe consider guarding the completions behind lib.optionalString (stdenvNoCC.buildPlatform.canExecute stdenvNoCC.hostPlatform) or similar.

[maxbrunet]

Thank you @SFrijters for giving this a try, I have added the canExecute condition

[katexochen]

Validated shell completion still works on x86_64-linux with bash, change looks good to me.
Didn't investigate further the cross-build issue.

[ulrikstrid]

Changes looks good, if @SFrijters can confirm that cross-builds works we can merge this.

[SFrijters]

Yep, it skips the completions, which is better than getting stuck completely.

[SFrijters]

FWIW, I looked into this a bit more and the problem seems to be that the register-python-argcomplete that is added to nativeBuildInputs is the host version, not the build version (in my case /nix/store/dfkxar40cwydm4dvap0v0093ws6klahy-python3.12-argcomplete-3.4.0-aarch64-unknown-linux-gnu/bin/register-python-argcomplete). If I replace py.pkgs.argcomplete -> buildPackages.python3Packages.argcomplete it finds /nix/store/k6pyxc68js2zrmqprg1w47i2xxk2mxl2-python3.12-argcomplete-3.4.0/bin/register-python-argcomplete and the completions are generated correctly in cross.

I don't understand why this is the case however; if I look at for example nixpkgs-review, which is also a Python package that uses register-python-argcomplete, just putting it into the nativeBuildInputs is enough. I assume something happens in the complex layering of the Azure packages in particular.