Optimized large SIMD butterflies for better register usage #286

Job	Run time
Check+Test no features (stable)	33s
Check and test WebAssembly with SIMD (stable)	1m 0s
Check+Test no features (beta)	37s
Check and test Linux arm 64bit with neon (stable)	1m 45s
Check+Test default features (stable)	1m 13s
Check and test WebAssembly with SIMD (beta)	1m 2s
Check+Test no features (nightly)	40s
Check and test Linux arm 64bit with neon (beta)	1m 54s
Check+Test default features (beta)	1m 18s
Check and test WebAssembly with SIMD (nightly)	1m 9s
Check+Test no features (1.61)	3m 15s
Check and test Linux arm 64bit with neon (nightly)	1m 47s
Check+Test default features (nightly)	1m 22s
Check and test WebAssembly with SIMD (1.61)	3m 15s
Check and test Linux arm 64bit with neon (1.61)	4m 51s
Check+Test default features (1.61)	3m 21s
Check and test Linux x86 32bit	1m 32s
Rustfmt	11s
	30m 45s

Provide feedback