As pointed out in issue #1472, currently the `run_sqr`

test doesn’t do anything with the result of the `fe_sqr`

call. Improve that by checking that the equation `(x+y)*(x-y) = x^2 - y^2`

holds for some random values y, as suggested by real-or-random in #1472 (comment). The existing loop for generating the x values is kept as-is, the sqr call on x is still included in each iteration on the course of calulating the RHS of the equation (`secp256k1_fe_sqr(&rhs, &x);`

).

The suggestion says “for random values `y != 0`

”, but I couldn’t see a good reason to exclude zero. Happy to apply s/random_fe_test/random_fe_non_zero_test/ if needed.