cmd/compile: add rules to use index regs for ppc64x

The following adds support for load and store instructions
with index registers, and adds rules to take advantage of
those instructions.

Examples of improvements:

crypto/rc4:
name     old time/op   new time/op   delta
RC4_128    445ns ± 0%    404ns ± 0%   -9.21%  (p=0.029 n=4+4)
RC4_1K    3.46µs ± 0%   3.13µs ± 0%   -9.29%  (p=0.029 n=4+4)
RC4_8K    27.0µs ± 0%   24.7µs ± 0%   -8.83%  (p=0.029 n=4+4)

crypto/des:
name         old time/op    new time/op    delta
Encrypt         276ns ± 0%     264ns ± 0%  -4.35%  (p=0.029 n=4+4)
Decrypt         278ns ± 0%     263ns ± 0%  -5.40%  (p=0.029 n=4+4)
TDESEncrypt     683ns ± 0%     645ns ± 0%  -5.56%  (p=0.029 n=4+4)
TDESDecrypt     684ns ± 0%     641ns ± 0%  -6.29%  (p=0.029 n=4+4)

crypto/sha1:
name          old time/op    new time/op    delta
Hash8Bytes       661ns ± 0%     635ns ± 0%  -3.93%  (p=1.000 n=1+1)
Hash320Bytes    2.70µs ± 0%    2.56µs ± 0%  -5.26%  (p=1.000 n=1+1)
Hash1K          7.14µs ± 0%    6.78µs ± 0%  -5.03%  (p=1.000 n=1+1)
Hash8K          52.1µs ± 0%    49.4µs ± 0%  -5.14%  (p=1.000 n=1+1)

Change-Id: I03810e90fcc20029975a323f06bfa086c973c2b0
Reviewed-on: https://go-review.googlesource.com/c/135975


Run-TryBot: Lynn Boger <laboger@linux.vnet.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Michael Munday <mike.munday@ibm.com>