Use logical shift right for BV '>>' operator

[JPaja]

I think operator >> should be implemented to use bvlshr as default since BitVectors dont have sign.

impl_binary_op!(
    BV<'ctx>,
    Shr,
    ShrAssign,
    shr,
    shr_assign,
    bvlshr,
    mk_const_bv
);

[waywardmonkeys]

@Pat-Lafon What do you think?

[Pat-Lafon]

The original comment might need a little bit more elaboration. The original pr #88 which added these operations specifically called out shr as the BV operation not to be added because of sign issues.

While BVs don't have a sign, the documentation I've read suggests that the theory relies on special operations that change whether the BV should be treated as having a sign.

If we are looking to have consistency with the python implementation, https://z3prover.github.io/api/html/classz3py_1_1_bit_vec_ref.html#a5977ca600992dbeda4a98b38c372133a shows >> as mapping to bvashr instead(the arithmetic version).

I'm not sure how surprising/unexpected this mapping is for users. For example https://stackoverflow.com/questions/13955385/z3py-do-a-roll#comment19249416_13955593 shows there is confusion... but if it makes sense for the python bindings then maybe not having >> implemented at all would be more confusing.