mwifiex: code rearrangement in multiport aggregation path

There are some macros defined for multiport aggregation
calculations. As we may need to add some more code to
accomodate new chipsets, we will change them to inline
functions. Also, use dynamic allocation for Rx buffer array.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>