lib.systems: Internally translate -gnuabielfv<1,2> to -gnu + gcc.abi

[OPNA2608]

While GCC supports the powerpc64-unknown-linux-gnuabielfv1 & powerpc64-unknown-linux-gnuabielfv2 triplets, Clang does not.
So using these triplets while configuring compilers and other tools that build upon them leads to issues and hacks.

Instead, set the name that should be used for -gnuabielfv<1,2> to just -gnu, with an annotation of the actual ABI for
GCC / the cc-wrapper. This still allows isAbiElfv<1,2> inspections to work without resulting in triplets that are
unknown to Clang & rustc.

---

I don't know if this is a good idea, I feel like it kinda just replaces hacks in one location with hacks in another… But it fixes some things ig… I don't feel like merging this as-is is fine without someone with a better understanding of & opinion on the lib.systems stuff having a look. Turning to draft once pings go out.

• powerpc64-unknown-linux-gnu triplet is still rejected by lib.systems.parse.abis.gnu.assertions - we want the user to be precise about which ABI they want. Maybe abit confusing cus the used triplet will be shown as powerpc64-unknown-linux-gnu, but passing that in will trigger an assertion failure.
• Change in lib/systems/examples.nix is because while passing those triplets as a string in crossSystem gets them parsed as expected, using pkgsCross.ppc64-elfv<1,2> just used the triplet as-is without any parsing of the components. Which lead to neither of the isAbiElfv<1,2> inspections working, amongst other issues.
• Untested natively on hardware, but cross doesn't seem to be going catastrophically wrong at least, and file of cross results reports the expected ABI version.
• Examples of issues / hacks that were necessitated by these triplets being given to compilers:
  • cc-wrapper: Workaround gnuabielfv{1,2} support deficiency on Clang #297425
  • lib.systems: Simplify ABI on POWER for rust.rustcTarget #451087
  • rust-bindgen/rustPlatform.bindgenHook: Missing target triplet (& ABI) fixes from CC wrapper #473862
  • pkgsCross.ppc64-elfv1.clangStdenv.cc: Can't find correct ld #490944
• Fixes pkgsCross.ppc64-elfv<1,2>.capnproto, and anything else cross that uses clangStdenv.

Things done

• Built on platform:
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• Tested, as applicable:
  • NixOS tests in nixos/tests.
  • Package tests at passthru.tests.
  • Tests in lib/tests or pkgs/test for functions and "core" functionality.
• Ran nixpkgs-review on this PR. See nixpkgs-review usage.
• Tested basic functionality of all binary files, usually in ./result/bin/.
• Nixpkgs Release Notes
  • Package update: when the change is major or breaking.
• NixOS Release Notes
  • Module addition: when adding a new NixOS module.
  • Module update: when the change is significant.
• Fits CONTRIBUTING.md, pkgs/README.md, maintainers/README.md and other READMEs.

[OPNA2608]

let
  pkgs = import ./. { };
  triplets = [
    "powerpc64-linux"
    "powerpc64-unknown-linux-gnuabielfv1"
    "powerpc64-unknown-linux-gnuabielfv2"
    "powerpc64-unknown-linux-musl"
  ];
  examples = [
    "ppc64-elfv1"
    "ppc64-elfv2"
  ];
  inheritDetails = platform: {
    inherit (platform) config isAbiElfv1 isAbiElfv2;
    parsed-abi = builtins.removeAttrs platform.parsed.abi [ "assertions" ];
    gcc-abi = platform.gcc.abi or null;
  };
in
(builtins.map (triplet:
  let
    crossPkgs = import ./. { crossSystem = triplet; };
  in
  (inheritDetails crossPkgs.stdenv.hostPlatform) // {
    via = "triplet-${triplet}";
  }
) triplets)
++ (builtins.map (example:
  let
    crossPkgs = pkgs.pkgsCross.${example};
  in
  (inheritDetails crossPkgs.stdenv.hostPlatform) // {
    via = "example-${example}";
  }
) examples)

[
  {
    config = "powerpc64-unknown-linux-gnu";
    gcc-abi = "elfv1";
    isAbiElfv1 = true;
    isAbiElfv2 = false;
    parsed-abi = {
      _type = "abi";
      abi = "elfv1";
      name = "gnu";
    };
    via = "triplet-powerpc64-linux";
  }
  {
    config = "powerpc64-unknown-linux-gnu";
    gcc-abi = "elfv1";
    isAbiElfv1 = true;
    isAbiElfv2 = false;
    parsed-abi = {
      _type = "abi";
      abi = "elfv1";
      name = "gnu";
    };
    via = "triplet-powerpc64-unknown-linux-gnuabielfv1";
  }
  {
    config = "powerpc64-unknown-linux-gnu";
    gcc-abi = "elfv2";
    isAbiElfv1 = false;
    isAbiElfv2 = true;
    parsed-abi = {
      _type = "abi";
      abi = "elfv2";
      name = "gnu";
    };
    via = "triplet-powerpc64-unknown-linux-gnuabielfv2";
  }
  {
    config = "powerpc64-unknown-linux-musl";
    gcc-abi = "elfv2";
    isAbiElfv1 = false;
    isAbiElfv2 = true;
    parsed-abi = {
      _type = "abi";
      name = "musl";
    };
    via = "triplet-powerpc64-unknown-linux-musl";
  }
  {
    config = "powerpc64-unknown-linux-gnu";
    gcc-abi = "elfv1";
    isAbiElfv1 = true;
    isAbiElfv2 = false;
    parsed-abi = {
      _type = "abi";
      abi = "elfv1";
      name = "gnu";
    };
    via = "example-ppc64-elfv1";
  }
  {
    config = "powerpc64-unknown-linux-gnu";
    gcc-abi = "elfv2";
    isAbiElfv1 = false;
    isAbiElfv2 = true;
    parsed-abi = {
      _type = "abi";
      abi = "elfv2";
      name = "gnu";
    };
    via = "example-ppc64-elfv2";
  }
]

[alyssais]

@Ericson2314 I remember discussing with you how we should handle mismatching triples between compilers with you years ago. What do you think is right here?