[WIP] Goldilocks

[MuhtasimTanmoy]

SIMD accelerated arithmetic and performance benchmark for SmallGoldilocks and its degree two extension field.

Function	Implementation	Test Working	Bench Working	Bench Result
arithmetic/add	✅	✅	-	-
arithmetic/sub	✅	✅	-	-
arithmetic/mul	✅	✅	-	-
evaluate_bf	✅	✅	✅	link
evaluate_ef	✅	✅	✅	link
reduce_bf	✅
reduce_ef	✅
sumcheck	✅

[z-tech]

Hi just getting around to this. This looks like clean code.

Overall what do the numbers look like comparatively ? I don't remember what the baseline numbers were. Do you have maybe a summary of the percent change (from the normal reduce /evaluate functions that take \F and the ones that take M31)?

I'm particularly interested in that detail about the branching logic and what difference (if any) it makes in performance.

Thanks and happy new (calendar) year!

[MuhtasimTanmoy]

@z-tech Happy new year! I will get back with the comparative summary.

[MuhtasimTanmoy]

Got this issue #90