LU-6261 gnilnd: Changes for small message rate improvment

LU-6261 gnilnd: Changes for small message rate improvment

Change number of threads on Aries service nodes to 7 for better
performance of small messages.
Change the max_immediate value to 8k instead of 2k. FMA has better
performance up to 8k.
lnet passes an niov of 256 reflecting the size of the router buffer
number of pages in the buffer pool when receiving data from the ib
interface when the message size is over one page. Adjust the niov to
reflect the message size/offset for sends.
Use the cookie defined in gni_pub.h instead of value from gnilnd.h.

Signed-off-by: Chris Horn <hornc@cray.com>
Change-Id: I84270653efcff56f06da7de7fb9674ae319800dd
Reviewed-on: http://review.whamcloud.com/15431
Tested-by: Jenkins
Tested-by: Maloo <hpdd-maloo@intel.com>
Reviewed-by: James Simmons <uja.ornl@yahoo.com>
Reviewed-by: Doug Oucharek <doug.s.oucharek@intel.com>
Reviewed-by: James Shimek <jshimek@cray.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>