Add Pixi workflow to verify Pyodide build works

[IvanIsCoding]

Follow up of #1447

This uses pixi to run the Pyodide build. This is still higly experimental. It is related to #1416 and #1420 in the sense that pixi is to conda what uv is to pypi.

The Github Action workflow I created does the following:

• Installs Python 3.12.7 from Conda via Pixi. Python needs to be pinned
• Installs Emscripten 3.1.58. This is also pinned. Newer versions are required to build with Python 3.13 but that is not on Conda at the moment.
• Installs Node.js
• Installs Rust 1.86. Generally we'd use nightly rust, but Conda does not offer that. So I had to use the RUSTC_BOOTSTRAP trick to enable nightly features on a compiler. Of course that is terrible, but the version is pinned so essentially it is like using a pinned nightly compiler
• Installs a pinned version of https://github.com/pyodide/pyodide-build
• The Pyodide CLI installs the tools to build for 0.27.7, which maps to Python 3.12
• Compiles the code with Rust and Emscripten

Tests are included in #1450

[coveralls]

Pull Request Test Coverage Report for Build 21499831342

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 2 unchanged lines in 1 file lost coverage.
• Overall coverage decreased (-0.01%) to 94.203%

---

Files with Coverage Reduction	New Missed Lines	%
rustworkx-core/src/generators/random_graph.rs	2	69.81%

Totals
Change from base Build 21464860520:	-0.01%
Covered Lines:	18283
Relevant Lines:	19408

---

💛 - Coveralls

[IvanIsCoding]

Emscripten 4.0.9 seems to be broken with binaryen, I'll need to investigate. It's a shame because I wanted to bump Pyodide such that we can test with the Pyodide shipped by uv

[IvanIsCoding]

Looks like uv might support a version compiled
with emscripten 3.1.58

[mtreinish]

Sorry this took me so long to follow up on. This looks good to me, it exciting to see a workflow for building pyodide in CI. I just had a couple of questions inline. I wasn't able to get pixi working locally and I'm wondering if something went stale while this was waiting on my review.

[mtreinish]

When I run this locally I first got an issue about the emcc -v call failing, when I set the env var to skip the version check it errors during the build with:

error: linking with `emcc` failed: exit status: 1
  |
  = note:  "emcc" "-s" "EXPORTED_FUNCTIONS=[\"_PyInit_generators\",\"_PyInit_rustworkx\"]" "<1 object files omitted>" "/home/mtreinish/git/qiskit/retworkx/target/wasm32-unknown-emscripten/release/deps/{libcompiler_builtins-ee68dc841fe2a394.rlib}.rlib" "-B<sysroot>/lib/rustlib/x86_64-unknown-linux-gnu/bin/gcc-ld" "--target=wasm32-unknown-emscripten" "-sDISABLE_EXCEPTION_CATCHING=0" "-o" "/home/mtreinish/git/qiskit/retworkx/target/wasm32-unknown-emscripten/release/deps/rustworkx.wasm" "-O3" "-g0" "-sSIDE_MODULE=2" "-sWASM_BIGINT" "-sSIDE_MODULE=2" "-sWASM_BIGINT" "-sABORTING_MALLOC=0" "-sWASM_BIGINT"
  = note: some arguments are omitted. use `--verbose` to show all linker arguments
  = note: Traceback (most recent call last):
            File "/home/mtreinish/git/qiskit/retworkx/.pixi/envs/default/bin/emcc", line 23, in <module>
              from tools.toolchain_profiler import ToolchainProfiler
          ModuleNotFoundError: No module named 'tools'

[IvanIsCoding]

So I have not been able to reproduce this. I tested on Linux locally, Mac locally, and on CI.

If you tell me the workaround you used, I can add it to pyproject.toml or at least mention it in CONTRIBUTING.md