docs: add benchmark results matrix to README

[waltsims]

Summary

Adds a ## Benchmarks section to the front-page README (docs/README.md) with a 10-row matrix covering (backend, OS, accelerator) combinations against three grid sizes: 64³, 128³, 256³.

All cells ship empty (em-dash) as a scaffold. The plan is for contributors to fill rows via follow-up commits on this branch:

• Magda: macOS (Apple Silicon), python backend, CPU
• Farid: any hardware you have available
• Anyone else with a machine that fits a row

Grid-size columns are drawn from the isotropic sizes in the default x/y/z_scale_array in benchmarks/helpers.py (indices 0/3/6/9 = 32³/64³/128³/256³, dropped 32³ as too small to be interesting). 512³ was excluded because many workstations OOM.

Rows omitted intentionally:

• python/cpp + macOS + GPU — no CUDA on macOS.
• cpp + Intel Mac + CPU — rare; can be added when someone runs it.

Reproducer command lives in benchmarks/README.md; the section links to it.

Test plan

• Section renders correctly on GitHub (table alignment, em-dash characters)
• Section renders correctly on PyPI once the next release ships (docs/README.md is the readme = in pyproject.toml)
• Contributors can fill a row without merge conflicts on other rows

🤖 Generated with Claude Code

Greptile Summary

This PR adds benchmark results to the public README. The main changes are:

• A backend, operating system, accelerator, and hardware matrix.
• Runtime columns for 64³, 128³, and 256³ grids.
• An initial Apple M1 result and contribution instructions.

Confidence Score: 5/5

The documentation is mergeable after small fixes to the benchmark links and reproduction guidance.

• The matrix dimensions and stated benchmark defaults match the implementation.
• The script link opens a different file than its label names.
• The default reproducer runs large cases that the matrix intentionally excludes.

docs/README.md

Important Files Changed

Filename	Overview
docs/README.md	Adds the benchmark matrix and contribution guidance; the benchmark defaults are accurate, but one link is mislabeled and the linked default reproducer runs cases outside the matrix.

_{Reviews (1): Last reviewed commit: "docs: fill python/macOS/CPU 64³ benchmar..." | Re-trigger Greptile}

Greptile also left 2 inline comments on this PR.

[waltsims]

@faridyagubbayli @djps

it would be great if you two could add entries here!

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 75.83%. Comparing base (dc21546) to head (9451b8a).
⚠️ Report is 3 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #769   +/-   ##
=======================================
  Coverage   75.83%   75.83%           
=======================================
  Files          58       58           
  Lines        8274     8274           
  Branches     1614     1614           
=======================================
  Hits         6275     6275           
  Misses       1373     1373           
  Partials      626      626           

Flag	Coverage Δ
3.10	75.80% <ø> (ø)
3.11	75.80% <ø> (ø)
3.12	75.80% <ø> (ø)
3.13	75.80% <ø> (ø)
macos-latest	75.71% <ø> (ø)
ubuntu-latest	75.71% <ø> (ø)
windows-latest	75.64% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[greptile-apps]

Benchmark Link Opens Wrong File

The link is labeled benchmarks/benchmark.py, but it opens benchmarks/README.md. Readers clicking it to inspect the benchmark implementation are sent to the instructions instead; the instructions already have a separate link below.

Suggested change

	Reference runtimes for the 3D scaling benchmark in [`benchmarks/benchmark.py`](../benchmarks/README.md). Values are total elapsed seconds for a single default run (3D initial-value problem, heterogeneous absorbing medium, 1000 timesteps, averaged over 3 repeats).
	Reference runtimes for the 3D scaling benchmark in [`benchmarks/benchmark.py`](../benchmarks/benchmark.py). Values are total elapsed seconds for a single default run (3D initial-value problem, heterogeneous absorbing medium, 1000 timesteps, averaged over 3 repeats).

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[greptile-apps]

Reproducer Runs Excluded Grid Sizes

The linked default command runs all 12 benchmark cases, including the large cases excluded from this matrix because workstations may run out of memory. A contributor following this instruction to fill a 64³, 128³, or 256³ cell can therefore spend time on unrelated cases or encounter an out-of-memory failure before the command completes; provide a bounded command that reproduces the listed sizes.