Simpler version of `to_euler_phases`

[moble]

Currently, to_euler_phases returns

zₐ ≔ exp(i*α)
zᵦ ≔ exp(i*β)
zᵧ ≔ exp(i*γ)

Presumably, the main consumer of this is SphericalFunctions's Wigner D and sYlm stuff, which exponentiate zₐ and zᵧ by $-m'$ and $-m$, respectively.

This is a bit more complicated than necessary; the more natural things to return are

z₊ ≔ exp(i*(α+γ)/2)
z₀ ≔ exp(i*β)
z₋ ≔ exp(i*(α-γ)/2)

Here, we would exponentiate z₊ and z₋ by $-m'-m$ and $-m'+m$, respectively. This would simplify half-integer cases a bit, but more importantly, I suspect that it might give slightly better results in some cases — possibly near β=π or 0.

I feel like that function should be called to_spinor_phases or something.