[GC] Fix & optimize arrayEquals on AArch64 (#743)

[weixlu]

[GC] Fix & optimize arrayEquals on AArch64 (#743)

Summary: fix tail comparison and simplify flow control for arrayEquals

Testing: jdk/javadoc/tool/parser/7091528/T7091528.java

Reviewers: mmyxym, sendaoYan

Issue: https://github.com/dragonwell-project/dragonwell11/issues/743

[weixlu]

About correctness cherry-pick this patch [Backport] 8139457: Relax alignment of array elements and the following test will fail

export test=test/langtools/jdk/javadoc/tool/parser/7091528/T7091528.java
function runJtreg() { jtreg -Xcomp -XX:+UseCompactObjectHeaders -ea -esa -timeoutFactor:4 -v:fail,error,time,nopass -nr -w $dir/index-$1 $test &> $dir/$1.log ; if [[ 0 -ne $? ]] ; then echo -n "$1 " ; else rm -rf $dir/index-$1 $dir/$1.log ; fi ; } ; export -f runJtreg ; export dir="tmp-jtreg-"`basename ${test##* } .java` ; rm -rf $dir ; mkdir -p $dir ; time seq 100 | xargs -i -n 1 -P `nproc` bash -c "runJtreg {}" ; echo total fail number: `ls $dir/*.log 2> /dev/null | wc | awk '{print $1}'`

with our fix, no failure is observed for five rounds of such test run.

[weixlu]

About performance	TestCase	Short	Medium	Long
testByteFalseBeginning	0.36%	17.58%	-4.66%
testByteFalseEnd	-24.83%	-6.15%	27.69%
testByteFalseMid	1.28%	9.20%	11.03%
testByteTrue	-29.56%	-7.45%	25.98%
testCharFalseBeginning	21.01%	-0.74%	-3.10%
testCharFalseEnd	-2.28%	30.00%	28.07%
testCharFalseMid	17.06%	13.15%	17.30%
testCharTrue	-0.18%	27.48%	28.67%

8.2% improvement on average.

There seems to be regression for short arrays. This is because the current implementation blindly loads a whole word, even if array doesn't fully occupy it. This is better for performance but introduces potential crash together with lilliput.

[weixlu]

This pr contains two commits, thus we decided to split it into two pr. New pr https://github.com/dragonwell-project/dragonwell11/pull/807 will track this fix