rt2x00: Add support for L2 padding during TX/RX

Some hardware require L2 padding between header and payload
because both must be aligned to a 4-byte boundary. This hardware
also is easier during the RX path since we no longer need to
move the entire payload but rather only the header to remove
the padding (mac80211 only wants the payload to be 4-byte aligned).

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>