[Fuzzing] Implement a Fp12 serializer to canonical serialization

[mratsim]

To fix https://github.com/guidovranken/el-fuzzers/issues/7 we need a serializer to canonical representation similar to BLST: supranational/blst#101.

This would allow merging into Geth fuzzer as well.

Note we have a towering switcher:

constantine/constantine/math/pairings/cyclotomic_subgroups.nim

Lines 675 to 702 in 1e34ec2

	func asFpk*[Fpkdiv6, Fpk](
	a: var Fpk,
	g0, g1: Fpkdiv6,
	g: G2345[Fpkdiv6]) =
	## Convert from a sextic extension to the Karabina g₂₃₄₅
	## representation.
	when a is CubicExt:
	when a.c0 is QuadraticExt:
	a.c0.c0 = g0
	a.c0.c1 = g1
	a.c1.c0 = g.g2
	a.c1.c1 = g.g3
	a.c2.c0 = g.g4
	a.c2.c1 = g.g5
	else:
	{.error: "a must be a sextic extension field".}
	elif a is QuadraticExt:
	when a.c0 is CubicExt:
	a.c0.c0 = g0
	a.c0.c1 = g.g4
	a.c0.c2 = g.g3
	a.c1.c0 = g.g2
	a.c1.c1 = g1
	a.c1.c2 = g.g5
	else:
	{.error: "a must be a sextic extension field".}
	else:
	{.error: "𝔽pᵏᐟ⁶ -> 𝔽pᵏ towering (direct sextic) is not implemented.".}

We might want to special-case this in a "fuzzing.nim" file as it might be that in the future Fp12 serialization is specified for the Whisk protocol that uses Fp12: https://ethresear.ch/t/the-return-of-torus-based-cryptography-whisk-and-curdleproof-in-the-target-group/16678

[mratsim]

Also check the discrepancy mentioned with libff:

• possibly change the cofactor clearing
• change the towering
• recheck vs Milagro/Miracl/AMCL which also use the same towering