add flake.nix for reproducible builds and NixOS

[randomizedcoder]

Add comprehensive Nix packaging support for PCP 7.0.5, enabling reproducible
builds via Nix flakes and preparing for inclusion in nixpkgs.

New files:

• flake.nix: Main Nix flake with package definition, dev shell, and VM test
• flake.lock: Pinned nixpkgs and flake-utils versions
• nix/vm-test.nix: NixOS VM integration test for pmcd daemon
• nix/patches/gnumakefile-nix-fixes.patch: Remove ownership flags, fix
    tmpfiles path, exclude qa from build
• nix/patches/tmpdir-portability.patch: Use ${TMPDIR:-/tmp} instead of
    hardcoded /var/tmp for sandbox compatibility
• nix/patches/configure-ar-portable.patch: Portable ar fallback (not
    applied, available for upstream)
• docs/HowTos/nix/index.rst: Comprehensive documentation explaining Nix
    packaging, technical details, and future improvements

Features enabled by default (Linux):

• Core PCP tools, libraries, and PMDAs
• Python3 and Perl language bindings
• BPF/BCC kernel tracing PMDAs
• Systemd integration
• SNMP, device mapper, and service discovery

The flake provides:

• nix build: Build PCP package with split outputs (out, man, doc)
• nix develop: Development shell with build dependencies and PCP logo
• nix build .#checks.x86_64-linux.vm-test: NixOS VM integration test

Technical fixes for Nix sandbox:

• Patch /var/tmp to ${TMPDIR:-/tmp} for sandbox compatibility
• Remove install ownership flags (handled by NixOS module)
• Fix broken symlinks with double /nix/store prefix
• Override systemd paths from pkg-config
• Export AR for configure script

Documentation ./docs/HowTos/nix/index.rst includes explanations of why Nix
packaging can be challenging (non-FHS paths, sandboxed builds, pure
environment) and how these constraints help expose portability issues in
build systems.

[kmcdonell]

@randomizedcoder Could I suggest you separate the 2 portability patches as a separate PR? I'm happy for those to proceed immediately.
Then you're left with just the one nix-specific patch and the associated nix life-support changes.
For the latter, my only (nit) issue is adding result* to the top-level .gitignore ... can that be more targeted, or does the nix build leave result* files all over the place?

[randomizedcoder]

G'day Ken,

Thanks for the feedback.

I'm actually from Melbourne also!

Is this what you meant? #2445

Thanks,
Dave