msm_rmnet_bam: changes to use netif_rx in place of netif_rx_ni

As per current implementation, we use netif_rx_ni for all
the packets which might give undue advantage to DL traffic
causing delays in UL traffic, especially TCP ACKs. This
results in tput drops. To avoid this, we can use netif_rx
in place of netif_rx_ni so that we process all the packets
in same context and there is no advantage for any data.
Also, we can invoke netif_rx_ni for every nth packet, this
value n can be configurable since it is defined as a module
parameter.

Change-Id: I413c2a284708b10b93bc176f36d039d4199a8065
Acked-by: Mohit Pahuja <mpahuja@qti.qualcomm.com>
Signed-off-by: Ravinder Konka <rkonka@codeaurora.org>