refactor: make clone_public_key infallible

[nicmr]

Hi, I was wondering why KeyPair::clone_public_key was fallible.
Turns out only the RsaPrivate -> RsaPublic conversion used TryFrom internally - a TryFrom implementation that never failed and just returned Ok, so it could just be From.
I've switched out TryFrom for From and switched the Result type for clone_public_key from Result<T> to T.

I've updated the tests and doctests touching clone_public_key to use the new signature so they all compile and pass again.

The only argument I see right now not to change the signature would be for forward compatibility for key types where there actually is a sensible error scenario for extracting the public key, which I find hard to imagine. Please let me know if I'm being ignorant here 🤔

Best Regards

[nicmr]

taking a look at the compile error

[nicmr]

Sorry, I totally missed to check with the openSSL backend, which is why this has to be fallible, and I only use the pure rust implementation locally. At least I understand the reason now. 🤦