Add key derivation function binding and appropriate tests

[mbestavros]

As the title suggests, this PR adds a binding for the PBKDF2-HMAC function in OpenSSL and configures it to match the implementation in Python-Keylime. An appropriate test is also added.

A few notes:

• This is also mentioned in the function documentation, but it bears repeating: The only reason this binding uses SHA-1 is because Python-Keylime uses it at the moment. PyCryptodome defaults to SHA-1 as a PRF if the prf parameter is not specified, and Python-Keylime does not provide that parameter in its KDF binding.

• The key the test compares against was generated from the Python-Keylime KDF binding using the same password and salt.

Tagging @frozencemetery and @leonjia0112 for review.

[leonjia0112]

@frozencemetery
I think you can try to delete all white space and type in space again. Otherwise, you can try rustfmt, that would solve this problem for you.

[mbestavros]

@frozencemetery: Made a few small style changes; CI build looks like it's working. cargo is complaining that the call to pbkdf2_hmac isn't handling errors correctly, though it's just a warning at the moment. Not sure of the best way to handle that, though.

[frozencemetery]

@mbestavros Basically its complaint is that pkcs5::pbkdf2_hmac has a return value which should be checked. How you do that depends a bit on the API you want to present; most likely, kdf() either needs to also return a Result<String>, or a failure brings down the system (unwrap or so).

[mbestavros]

@frozencemetery: Got it; per our offline discussion, I made kdf and do_hmac pass the Result returned from OpenSSL down the line until it can be appropriately handled. Also made appropriate changes to main code to make sure the new function calls work. Figured the slight context-switch deserved its own commit.

Let me know what you think.

[frozencemetery]

Looks good. Squash the changes to kdf() into a single commit and shorten the second commit message, then it's ready to go.

[mbestavros]

@frozencemetery: Done, commits are properly separated now.

[frozencemetery]

Okay, so this is better in that we're not panicing in the crypto layer, but this will still cause the server to panic if it fails. Can you just fail the request instead?

[mbestavros]

@frozencemetery Fixed; the call to HMAC from within main.rs now should fail gracefully. I used the match pattern; if there's a better way, please let me know.

[frozencemetery]

Great, thanks!