fix: better accuracy lanczos fourier-space interp

[beckermr]

jax-galsim uses a lower-accuracy akima interpolant instead of a cubic spline when interpolating the k-space table for the fourier transform of the Lanczos interpolant. This causes a loss in accuracy. For jax-galsim, I am increasing the number of points being interpolated by 2.3x. The cost of computing the table comes at compile time, and it is cached on the CPU side, so we should not see big performance losses in most cases.

xref: GalSim-developers/GalSim-for-JAX-GalSim-Testing#9

[codspeed-hq]

Merging this PR will degrade performance by 33.32%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

❌ 1 regressed benchmark
✅ 35 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	test_benchmarks_lanczos_interp[kval-no_conserve_dc-run]	291.2 µs	436.8 µs	-33.32%

---

_{Comparing test-fixes-4 (0bf5eee) with main (1e20520)}