[experiment] parallell trie hashing

[holiman]

This is a very dirty attempt to see how much faster it would be to hash the trie if we use some more goroutines. This poc uses four threads, and becomes ~35% faster (than #20461).

[user@work trie]$ benchstat noparalell_hash.txt parallell_4_hash.txt 
name    old time/op    new time/op    delta
Hash-6    4.34µs ±12%    2.82µs ± 7%  -35.06%  (p=0.000 n=10+9)

name    old alloc/op   new alloc/op   delta
Hash-6      673B ± 0%      683B ± 0%   +1.37%  (p=0.000 n=9+9)

name    old allocs/op  new allocs/op  delta
Hash-6      9.00 ± 0%     10.00 ± 0%  +11.11%  (p=0.000 n=10+10)

Some early attempt showed roughly similar figures for two threads, so there doesn't seem to be much gain in further increasing the threads.

And yes, the code is butt ugly, and also breaks some other usecases (like, creating proofs). Do we think this is worth pursuing? If so, I can make a stab at doing it properly.

[holiman]

Closing in favour of #20488