perf: Add the LegacyMultiMut , and investigate where its applicable.

[ValarDragon]

Summary

In #7762 I introduced a method I called "legacyMultiMut", which would significantly speedup even multiplication of two Dec's, by not needing to use QuoRem and instead just not divide out the final factor of two, then look at the final bit, and do a final rsh.

This should imply faster multiplication of Dec and BigDec throughout our codebase, if we can tolerate the rounding being slightly different than what we do today.

I put this on halt due to other priorities, but it would be good to at least get chopPrecisionAndRoundParity method exposed.

If we find ourselves bottlenecked on Dec operations, this is a good thing to investigate.

Problem Definition

.

Proposed Feature

.