srp: compute `K = H(S)` correctly.

[zadlg]

srp: compute K = H(S) correctly.

According to the specification section The SRP protocol, K (which corresponds to the session key) is computed as follows:

K = H(S)

where H is the digest algorithm, and S is the common exponential value.

In the current implementation, K is equal to S, which does not follow the SRP protocol specification.

This commit fixes this issue by computing the right value for K.

[tarcieri]

See also #153 and #163, as well as RFC5054

[zadlg]

up

[tarcieri]

@zadlg your opinion on the existing PRs and any overlap and existing discussion would be appreciated

[zadlg]

Sorry, I didn't understand you were waiting for an answer from me.

I think the specification is quite clear, isn't it ?

[tarcieri]

Let me be a little more explicit:

• Do you think your PR is a duplicate?
• What do you think about the approach in the other PR?
• Which PRs should be merged?