[Eth2.0] Modular_squareroot issue due to upstream Py-ECC bug

[mratsim]

What is wrong?

FYI due to ethereum/py_ecc#47 with impact details in ethereum/consensus-specs#508, the modular squareroot returned does not always match the one expected by the Eth2.0 spec.

trinity/eth2/_utils/bls.py

Lines 63 to 77 in 8688cf7

	def modular_squareroot(value: FQ2) -> FQP:
	"""
	``modular_squareroot(x)`` returns the value ``y`` such that ``y**2 % q == x``,
	and None if this is not possible. In cases where there are two solutions,
	the value with higher imaginary component is favored;
	if both solutions have equal imaginary component the value with higher real
	component is favored.
	"""
	candidate_squareroot = value ** ((FQ2_order + 8) // 16)
	check = candidate_squareroot ** 2 / value
	if check in eighth_roots_of_unity[::2]:
	x1 = candidate_squareroot / eighth_roots_of_unity[eighth_roots_of_unity.index(check) // 2]
	x2 = FQ2([-x1.coeffs[0], -x1.coeffs[1]]) # x2 = -x1
	return x1 if (x1.coeffs[1], x1.coeffs[0]) > (x2.coeffs[1], x2.coeffs[0]) else x2
	return None

How can it be fixed

Wait for upstream fix for ethereum/py_ecc#47