fix(analysis/inner_product_space,geometry/euclidean): two deterministic timeout fixes (#16365)
CI seems to be having some issues after #16356, which I can't reproduce with `-T100000` locally but can with `-T90000`. This PR speeds up `analysis/inner_product_space/l2_space.lean:hilbert_basis.coe_mk` (which was already investigated before in #15271) and `geometry/euclidean/oriented_angle.lean:inner_eq_norm_mul_norm_mul_cos_oangle`.