Add support for C contiguous arrays

[pavanky]

Many python modules default to C contiguous layouts. Adding support for this would be nice.

This can be solved independently of upstream. Most functions are independent of the layout. The few that are dependent include the following.

• Functions that depend on the dimension number: reductions / sort / scan / ffts.
• Linear algebra functions that assume column major layout: matmul, lu, qr, cholesky, solve.
• Perhaps some others I am missing.