cmd/compile: use generated loops instead of DUFFCOPY on riscv64

cmd/compile: use generated loops instead of DUFFCOPY on riscv64

MemmoveKnownSize112-4            632.1Mi ± 1%   1288.5Mi ± 0%  +103.85% (p=0.000 n=10)
MemmoveKnownSize128-4            636.1Mi ± 0%   1280.9Mi ± 1%  +101.36% (p=0.000 n=10)
MemmoveKnownSize192-4            645.3Mi ± 0%   1306.9Mi ± 1%  +102.53% (p=0.000 n=10)
MemmoveKnownSize248-4            650.2Mi ± 2%   1312.5Mi ± 1%  +101.87% (p=0.000 n=10)
MemmoveKnownSize256-4            650.7Mi ± 0%   1303.6Mi ± 1%  +100.33% (p=0.000 n=10)
MemmoveKnownSize512-4            658.2Mi ± 1%   1293.9Mi ± 0%   +96.60% (p=0.000 n=10)
MemmoveKnownSize1024-4           662.1Mi ± 0%   1312.6Mi ± 0%   +98.26% (p=0.000 n=10)

Change-Id: I43681ca029880025558b33ddc4295da3947c9b28
Reviewed-on: https://go-review.googlesource.com/c/go/+/700537
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Reviewed-by: Mark Freeman <markfreeman@google.com>