U256 div measurement

[rakita]

Currently doing some measurements on revm and how division is done in U256.
link here: bluealloy/revm#7 (comment)

In short, I am getting a performance boost from ~77-80ms to ~64-66ms
by using this part of code:

parity-common/uint/src/uint.rs

Lines 1181 to 1185 in 5fe7924

	// NOTE: this is slow (__udivti3)
	// let x = (u128::from(hi) << 64) + u128::from(lo);
	// let d = u128::from(d);
	// ((x / d) as u64, (x % d) as u64)
	// TODO: look at https://gmplib.org/~tege/division-paper.pdf

[ordian]

See #406, #478. TL;DR is the benches result were inconsistent last time we run them. Would be good to dig into this again.

[ordian]

closed by #478