Properly cast arrays before calling C code

[tmcclintock]

This is actually two issues.

1. Arrays in Python can sometimes not be C-ordered and require a numpy.ascontiguousarray() call. This can be handled internally in the toolkit.

2. Similarly, the C code in the toolkit assumes double precision (float64). If a numpy array is floating point precision (float32) then interpolation errors can occur. This can be handled with proper casting at the same time that issue (1) is handled.