nvme-pci: optimize mapping of small single segment requests

If a request is single segment and fits into one or two PRP entries we
do not have to create a scatterlist for it, but can just map the bio_vec
directly.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>