v2(nightly) - Adding Fp128 implementation based on fiat-crypto

[armfazh]

fiat_crypto without primitives

This is an implementation of Fp128 using fiat-crypto. The code generated needs a wrapper to implement the traits required by FieldElement.

The downside of this implementation is the performance observed. To improve on top of #758 , this PR implements the primitives using rust intrinsics available only in nightly.

Timings

Fp128		Primitives		No Primitives
poly_mul/fft	Current	fiat-crypto	Change	fiat-crypto	Change
1	180.14 ns	281.47 ns	+57.701%	237.39 ns	+56.250%
30	5.7912 µs	10.398 µs	+79.487%	8.8421 µs	+52.681%
60	12.925 µs	24.852 µs	+87.113%	20.228 µs	+56.502%
90	29.935 µs	53.975 µs	+80.685%	47.558 µs	+58.870%
120	28.688 µs	59.180 µs	+101.55%	46.196 µs	+61.029%
150	64.779 µs	127.07 µs	+97.426%	104.41 µs	+61.178%