Use OpenSSL's memory allocator

[ghost]

OpenSSL includes a replaceable memory allocator. If we decide to simulate failures using an instrumented allocator for #9, it makes sense to use OpenSSL's functions so that we don't need to write our own version of CRYPTO_set_mem_functions.

On the other hand, this makes things harder for callers, who would need to import openssl/crypto.h to free buffer contents. I'm also not sure how well this would work if we add an NSS backend later.

[ghost]

This is easy to do now that we're not going to return allocated buffers.

[ghost]

We only use malloc in the header parser, which we'll remove in #28, and calloc in ece_compute_secret. I don't think this is worth doing.