@hackage reinterpret-cast0.1.0

Memory reinterpretation casts for Float/Double and Word32/Word64.

The implementations in the .Internal package are different ways to tackle the problem; the array method (current default) is about 5 times faster than the FFI method.