perf: try using fori_loop but unrolling everything

[beckermr]

I am seeing if this has the same performance as the manual loops.

[codspeed-hq]

CodSpeed Performance Report

Merging #141 will not alter performance

_{Comparing fori_loop_unroll (2dac00e) with main (3e45fb5)}

Summary

✅ 29 untouched benchmarks
🆕 2 new benchmarks

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
🆕	test_benchmark_spergel_calcfluxrad[compile]	N/A	15.4 s	N/A
🆕	test_benchmark_spergel_calcfluxrad[run]	N/A	2.1 ms	N/A