First pass at implementation/tests for MR with least squares

As you'll see, some of the tests - namely all of the 2nd-order tests with extended histories - are failing. I'm still not sure why this is. Having used test_singlerate_squarewave as a base example, I'm finding that I can't even get it to fail regularly, so I think it might be an issue with something being uninitialized in the Fortran builtin stuff.

There are also a few 5th order tests failing less catastrophically, and I'm less sure what to do about those...