SHA3: changing java implementation to help C2 create high-performance code

SHA3 algorithm iteratively performs arithmetic operations on a batch of 25 long values, see SHA3.keccak function: https://github.com/openjdk/jdk/blob/master/src/java.base/share/classes/sun/security/provider/SHA3.java#L258

Manual inlining, manual unrolling, and moving the data from array[25] into local variables speed up code execution up to 2 times.

    // see the attached diff
    private void keccak() { // ARM64 AMD
        keccak_0_default(); // 2764ms 1846ms
        // keccak_1_nativeimpl(); // 1701ms 1649ms
        // keccak_2_inlined_unrolled(); // 1754ms 1499ms
        // keccak_3_inlined_unrolled_localvars(); // 1437ms 1261ms
    }