chore(nix): update flake inputs (nixpkgs 2026-04 → 2026-07)

[alt-glitch]

Summary

Updates all flake inputs to latest (~3.5 months of drift):

input	old	new
nixpkgs (nixos-unstable)	2026-04-01 (6201e203)	2026-07-19 (241313f4)
uv2nix	2026-03-09	2026-07-20
pyproject-nix	2026-03-07	2026-07-20
pyproject-build-systems	2026-03-03	2026-07-20
flake-parts	2026-03-01	2026-07-01
npm-lockfile-fix	unchanged	(no newer commit)

Stacked on #69458 (base branch is fix/desktop-electron-headers; retarget to main after it merges). The nixpkgs bump moves Electron 41.0.2 → 41.9.1, which trips the hardcoded headers hash #69458 removes — the only breakage the entire input update causes.

Test plan (all on new pins, x86_64-linux)

• nix build .#default — green (venv, tui, web rebuilt cleanly under new uv2nix/nixpkgs)
• nix build .#desktop — green with fix(nix): use electron.headers instead of hardcoded headers hash in desktop.nix #69458 (reproduces the nix: .#desktop build breaks on every nixpkgs electron bump (hardcoded node-headers hash) #61443 hash mismatch without it)
• nix flake check -L — all 16 checks pass

[github-actions]

૮ >ﻌ< ა ci review

_{running on 76ec8c3}

CI timings

CI timings · View job

Wall time 8m21s vs 11m30s (-27.4%). 11 job(s) slower, 8 faster, 1 unchanged.

• Build&Test Docker image / build (amd64, ubuntu-latest, linux/amd64, type=gha,scope=docker-amd64, type=gha,mode=max,scope=do...: -184.0s
• Python tests / Run tests slice 5/8: +77.0s
• Python tests / Run tests slice 6/8: +34.0s
• Python tests / Run tests slice 3/8: -33.0s
• Build&Test Docker image / build (arm64, ubuntu-24.04-arm, linux/arm64, type=gha,scope=docker-arm64, type=gha,mode=max,scope...: -33.0s

[alt-glitch]

Call-graph-assisted review — no compiled code changes here (flake.lock + Nix), so calldiff has nothing to diff; reviewed via git diff against the merge-base.

Straightforward flake input bump: flake-parts, nixpkgs, pyproject-build-systems, pyproject-nix, and uv2nix all advanced, and the duplicated input aliases pyproject-nix_2 / uv2nix_2 are renamed back to pyproject-nix / uv2nix. The alias de-duplication is a nice cleanup.

🟡 Duplication / abstraction

• This PR carries the exact same nix/desktop.nix change as fix(nix): use electron.headers instead of hardcoded headers hash in desktop.nix #69458 (drop the hardcoded electronHeaders fetchurl, use ${electron.headers}). Since this branch is stacked on fix(nix): use electron.headers instead of hardcoded headers hash in desktop.nix #69458's base, that's expected, but flag it explicitly: land fix(nix): use electron.headers instead of hardcoded headers hash in desktop.nix #69458 first, then this so the shared hunk doesn't produce a spurious conflict or a double-review of the same change. If fix(nix): use electron.headers instead of hardcoded headers hash in desktop.nix #69458 is dropped, this diff would need the desktop.nix hunk re-justified on its own.

🔴 Issues

• Nothing blocking. One thing to confirm before merge: nix flake check / a desktop build actually pass against the new nixpkgs rev (241313f…) — input bumps of this size occasionally shift electron/Python package versions in ways that ripple into desktop.nix and the uv2nix set. The trigger-only Nix CI from ci(nix): restore nix CI as trigger-only workflow with local entrypoint #69463 (label ci/nix) is the right gate to run here.

✅ Looks good

• Collapsing pyproject-nix_2→pyproject-nix and uv2nix_2→uv2nix removes redundant lock nodes and makes the follows-graph read cleanly.
• contributors/emails/... credit entry is consistent with the salvage note referenced by fix(nix): use electron.headers instead of hardcoded headers hash in desktop.nix #69458.