shake128 test coverage about sha.jl

[immoschuett]

Hi, thanks for the tests @inkydragon

I also thought about the padding.

I think the same issue is also in the digest for the other SHA3 hash functions.
The padding of julia SHA is not along the NIST standards.

the Padding in the else case usedspace == blocklen(T)-1 should be

context.buffer[end] = 0x9f for SHAKE_CTX

I tested the new padding and the output matches the python output then.

and one can also change the padding in the digest for SHA3_CTX

context.buffer[end] = 0x86 for sha3 digest function, but this (probably) is ivariant in the transform. So one would not need transform twice

[inkydragon]

Fixed by #95