Implement a ->digest function for sha256-ssse3, sha256-avx, sha256-avx2,
and sha256-ni.  This improves the performance of crypto_shash_digest()
with these algorithms by reducing the number of indirect calls that are
made.

For now, don't bother with this for sha224, since sha224 is rarely used.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>