RDMA/siw: Move tx_cpu ahead

We can reduce one cacheline for the usage of struct siw_qp.

Before,

	/* size: 1928, cachelines: 31, members: 38 */
	/* sum members: 1920, holes: 2, sum holes: 8 */
	/* paddings: 4, sum paddings: 13 */
	/* forced alignments: 3 */

after

	/* size: 1920, cachelines: 30, members: 38 */
	/* paddings: 4, sum paddings: 13 */
	/* forced alignments: 3 */

Link: https://lore.kernel.org/r/20231203092655.28102-2-guoqing.jiang@linux.devAcked-by: Bernard Metzler <bmt@zurich.ibm.com>
Signed-off-by: Guoqing Jiang <guoqing.jiang@linux.dev>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

drivers/infiniband/sw/siw/siw.h

@@ -417,10 +417,10 @@ struct siw_iwarp_tx {
 struct siw_qp {
 	struct ib_qp base_qp;
 	struct siw_device *sdev;
+	int tx_cpu;
 	struct kref ref;
 	struct completion qp_free;
 	struct list_head devq;
-	int tx_cpu;
 	struct siw_qp_attrs attrs;
 
 	struct siw_cep *cep;