ring-math: use native integers for polynomial norms when possible

[chancehudson]

If polynomial_degree * scalar_prime < 2^64 it should be safe to use e.g. a u64 instead of a BigUint

[chancehudson]

Related, instead of returning the norm value we could expose a function that returns a boolean. Something like

l_max_norm_lte(&self, upper_bound: u64): this returns a boolean indicating if the l2 norm is less than or equal to the upper bound. This allows the function to exit early and potentially use smaller integer representations.