implement slicing-by-2

[fracai]

I came across a reference to a "slicing-by-2" technique, but I haven't seen this actually implemented anywhere.

This PR implements that algorithm and the speed slots right in between the single byte and slicing-by-4 variants.

On my system:

$ ./Crc32Test 
Please wait ...
bitwise          : CRC=221F390F, 8.717s, 117.475 MB/s
half-byte        : CRC=221F390F, 5.740s, 178.384 MB/s
tableless (byte) : CRC=221F390F, 6.606s, 155.019 MB/s
tableless (byte2): CRC=221F390F, 9.292s, 110.207 MB/s
  1 byte  at once: CRC=221F390F, 2.707s, 378.304 MB/s
  2 bytes at once: CRC=221F390F, 1.687s, 606.836 MB/s
  4 bytes at once: CRC=221F390F, 0.969s, 1056.677 MB/s
  8 bytes at once: CRC=221F390F, 0.535s, 1914.827 MB/s
4x8 bytes at once: CRC=221F390F, 0.536s, 1909.030 MB/s
 16 bytes at once: CRC=221F390F, 0.281s, 3650.535 MB/s
 16 bytes at once: CRC=221F390F, 0.280s, 3651.705 MB/s (including prefetching)
    chunked      : CRC=221F390F, 0.281s, 3644.816 MB/s

I don't have a way to test the Arduino code, but it "looks right to me".