get_LMs in C++

[bdice]

fsph should include a C++ function equivalent to the Python function get_LMs that it exposes. It would make freud's indexing w.r.t. spherical harmonics much more transparent if fsph actually told you the order in which it returned spherical harmonics, ideally via a simple iterator.

See also: glotzerlab/freud#371

[klarh]

I would be OK with this. We probably want three variants (which could simply be functions of the iterator): one for l, one for m (signed), and one for m_index (unsigned).