Introduce standalone gost_digest adaptor for provider

[Mironenko]

This PR addresses the digest-related part of issue #505.

It introduces a new family of gost_digest* source files implementing an adapter layer over gosthash and gosthash2012.

The adapter is used by gost_prov_digest and removes its dependency on engine-specific code.

In the future, the gost_digest* implementation may also replace gost_md / gost_md2012 in the engine backend.

This PR also fixes test_digest to correctly test the provider implementation. Prior to this change, test_digest invoked the engine-based digest implementation in both test-digest-engine and test-digest-provider, so the provider code path was not actually tested.

[chipitsine]

does not look severe, but ...

*** CID 644776:         Resource leaks  (RESOURCE_LEAK)
/test_digest.c: 896             in do_synthetic_once()
890     
891         EVP_MD *dgst;
892         T((dgst = EVP_MD_fetch(NULL, tv->algname, NULL))
893           || (dgst = (EVP_MD *)EVP_get_digestbyname(tv->algname)));
894         if (dgst && warn_md_impl_is_expected(dgst)) {
895             EVP_MD_free(dgst);
>>>     CID 644776:         Resource leaks  (RESOURCE_LEAK)
>>>     Variable "ibuf" going out of scope leaks the storage it points to.
896             return 1;
897         }
898     
899         OPENSSL_assert(EVP_MD_is_a(dgst, tv->algname));
900         EVP_MD_CTX *ctx, *ctx2;
901         T(ctx  = EVP_MD_CTX_new());