chore: bump cpu/cuda pins to the RLS/JRLS kernel fixes and test coverage

[balbasty]

Final link in the propagation chain from an independent re-review of the fastfields-cpu-lib#65 RLS/JRLS fix chain.

cuda: b8e5e9c → 6ae52bd (fastfields-cuda-lib#46)

Completes propagation of fastfields-cuda-impl#41, which moved cuda-impl's kernels pin to main after it had fallen nine commits behind. The CUDA path was missing two field/flow regulariser correctness fixes the CPU path already had:

fix	effect
fastfields-kernels#64	make_kernel_bending_rls rescaled its whole coefficient table by 0.25, running the membrane penalty at half strength whenever the bending-order RLS/JRLS kernel is built (bending != 0 alongside membrane != 0). Upstream measured max |matvec_bending - matvec_bending_rls(w=1)| of 1.8e+00 (2D) / 1.7e+00 (3D) before, ~1e-14 after. The existing self-adjointness tests are blind to it — a term at the wrong strength stays perfectly symmetric.
fastfields-kernels#52	guards the weight-map neighbour reads in the RLS/JRLS kernels (out-of-bounds read at some boundaries).

With this, the CPU and CUDA paths finally sit on the same kernels commit (b09b284).

cpu: bf57a02 → 13af367 (fastfields-cpu-lib#83)

Adds ground-truth coverage for the wc == 1 (JRLS broadcast) direction, which previously had none. The #65 predicate bug mis-routed both weight-map layouts, but only wc == nc was regression-tested; the wc == 1 case went to the per-channel _rls kernels and read out of bounds past the single weight channel, yet every test exercising it stayed green because each compared the operator against itself.

Measured on that branch, reverting only #73's predicate flip: 18284 checks / 6141 failures, of which 3056 come from the newly-covered direction (previously 0). Clean run: 18284 / 0.

Also unswaps the RLS/JRLS labels in cpu-lib's public header and test comments.

Verification

No source changes here. Both sub-libraries' own CI is green on the exact pinned commits, including fastfields-cuda-lib#46's 51-minute nvcc build.

🤖 Generated with Claude Code

https://claude.ai/code/session_016AjQcY78NgbagPSbPJRr6Z

---

Generated by Claude Code

[github-actions]

Submodule staleness (non-blocking -- fastfields-lib#15)

submodule	pinned commit	tracks	status	behind by
cpu	13af367	fastfields/fastfields-cpu-lib@main	up to date	0
cuda	6ae52bd	fastfields/fastfields-cuda-lib@main	up to date	0