Reorg + Spiral Explorer + Perf

[jacobrosenthal]

There were a few allocations in there so thought I could beat you. Didnt do anything much on my end sadly though. Its prettier and more functional though with more deps :(

Feel free to close, but give it a shot on your high core machine and let me know if it eeks out any extra hashrate.

[jacobrosenthal]

with my par iter across the tuple pairs i think I make far more jobs for rayon, presumably better if you have many threads and huge blocks but actually a little worse for small core count machines

my poor little 8core xps13

upstream algo (bench_pure_them)
chunkLength 16: 858.1964465303386
chunkLength 32: 687.9408800806181
chunkLength 64: 581.5208132205122

upstream algo on my allocation and debug assert changes (bench_them)
chunkLength 16: 981.5950920245398
chunkLength 32: 891.98606271777
chunkLength 64: 613.1002275176625

my algo and allocation and debug assert changes (bench)
chunkLength 16: 899.1921320688443
chunkLength 32: 600.4456432508503
chunkLength 64: 604.6559690586204

It preforms only comparable to node headless miner we ripped out of the client which seems suprising. I know js is optimized, but..
~800 at 16 size

[jacobrosenthal]

Theres still some loss in here and I just cant see it..
On a digital ocean $5 droplet (1 vcpu) the node version does ~326

But basically all 3 of the benchmarks I have here do ~238, or ~180 with rayon removed (so its not that)...