StringBuilder::compareTo performs worse than String::compareTo

A DESCRIPTION OF THE PROBLEM :
The reason is that StringBuilder::compareTo ends up calling
StringLatin1::compareTo(byte[] value, byte[] other, int len1, int len2)
Which is not intrinsified nor vectorized.
String::comapreTo ends up calling StringLatin1::compareTo(byte[] value, byte[] other) which is intrinsified and vectorized.

A simple solution can be to replace the code in
StringLatin1::compareTo(byte[] value, byte[] other, int len1, int len2)
with Arrays::compare.

Below are benchmark results comparing String::compareTo, StringBuilder::compareTo and Arrays::compare.

####### TEST CODE #######

@Fork(2)
public class CompareToBenchmark {
    private static final String string1 = new String(new byte[1023]);
    private static final String string2 = new String(new byte[1023]) + "a";
    private static final StringBuilder stringBuilder1 = new StringBuilder(string1);
    private static final StringBuilder stringBuilder2 = new StringBuilder(string2);
    private static final byte[] bytes1 = string1.getBytes();
    private static final byte[] bytes2 = string2.getBytes();

    @Benchmark
    @BenchmarkMode(Mode.AverageTime)
    @OutputTimeUnit(TimeUnit.NANOSECONDS)
    public int compareStrings() {
        return string1.compareTo(string2);
    }

    @Benchmark
    @BenchmarkMode(Mode.AverageTime)
    @OutputTimeUnit(TimeUnit.NANOSECONDS)
    public int compareStringBuilders() {
        return stringBuilder1.compareTo(stringBuilder2);
    }

    @Benchmark
    @BenchmarkMode(Mode.AverageTime)
    @OutputTimeUnit(TimeUnit.NANOSECONDS)
    public int compareByteArrays() {
        return Arrays.compare(bytes1,0, bytes1.length, bytes2, 0, bytes2.length);
    }
}

####### RESULTS #######

Benchmark Mode Cnt Score Error Units
CompareToBenchmark.compareByteArrays avgt 10 32.179 ± 5.666 ns/op
CompareToBenchmark.compareStringBuilders avgt 10 353.890 ± 115.107 ns/op
CompareToBenchmark.compareStrings avgt 10 29.072 ± 1.037 ns/op