Add support for the non-transposed RHS matrix in the packing function

- Remove nxk reference into the original packing function
- Add input argument in the packing function to tell whether
the RHS matrix is N x K or K x N
- Extend the example to work with N x K or K x N RHS matrices
- Expose in the header file of the packing function the RHS packed stride

Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>