There are two types of basemuls in ML-DSA we need to implement:

• A vector-vector inner product (aka polyvecl_pointwise_acc_montgomery) - this will be specialized for L=4,5,7 similar to the basemul in mlkem-native (this issue)
• A vector-scalar product (aka polyveck_pointwise_poly_montgomery) - this will be generic, i.e., we will put the poly_pointwise_montgomery in asm (see Add Neon poly_pointwise_montgomery #207)

Unfortunately, we don't currently have code that can simply be pulled in.