runtime: remove AVX2less code in memclrNoHeapPointers for GOAMD64 >= 3

runtime: remove AVX2less code in memclrNoHeapPointers for GOAMD64 >= 3

Optimize memclr by removing simple case loop along with the
runtime check since AVX2 is guaranteed to be available when
compiling with GOAMD64 >= 3

name             old speed      new speed      delta
Memclr/5-12      2.70GB/s ± 1%  2.73GB/s ± 1%     ~     (p=0.056 n=5+5)
Memclr/16-12     7.00GB/s ± 2%  7.03GB/s ± 1%     ~     (p=1.000 n=5+5)
Memclr/64-12     25.5GB/s ± 1%  25.5GB/s ± 1%     ~     (p=0.548 n=5+5)
Memclr/256-12    53.4GB/s ± 1%  52.7GB/s ± 2%     ~     (p=0.222 n=5+5)
Memclr/4096-12    109GB/s ± 1%   129GB/s ± 0%  +18.57%  (p=0.008 n=5+5)
Memclr/65536-12  75.2GB/s ± 2%  78.3GB/s ± 3%   +4.14%  (p=0.008 n=5+5)
Memclr/1M-12     53.5GB/s ± 2%  54.1GB/s ± 2%     ~     (p=0.310 n=5+5)
Memclr/4M-12     53.1GB/s ± 3%  52.9GB/s ± 2%     ~     (p=1.000 n=5+5)
Memclr/8M-12     44.6GB/s ± 3%  45.1GB/s ± 3%     ~     (p=0.310 n=5+5)
Memclr/16M-12    24.8GB/s ± 2%  24.2GB/s ± 2%     ~     (p=0.056 n=5+5)
Memclr/64M-12    38.3GB/s ± 1%  37.8GB/s ± 1%     ~     (p=0.056 n=5+5)
[Geo mean]       31.0GB/s       31.5GB/s        +1.78%

Change-Id: I6f3014f6338cb3b5a1b94503faa205f043fe2de8
Reviewed-on: https://go-review.googlesource.com/c/go/+/367494
Trust: Cherry Mui <cherryyz@google.com>
Trust: Daniel Martí <mvdan@mvdan.cc>
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Keith Randall <khr@golang.org>