More various fixes (F-*)

[gasbytes]

eca9700 - Added Ed25519KeyPair to prevent type_complexity warning (cargo clippy) from flagging and to make it more readable
29a7fba - cargo fmt formatting
7e494c4 - remove unnecessary wc_ecc_curve_id call
e5bd0e0 - check the digest size after getting its size from the hash type
e3484f4 - remove unnecessary bitwise copy of RsaKey(s) and its deferencing, use size_of the type directly instead
6ad05ae - move test related crates from dependencies to dev-dependencies instead since they are not used in the library
5380da7 - propage init() failures to callers instead of panicking, so a failed wc_init* surfaces as a retriaviable error
2c979db - zeroize ecdsa exported scalar in case of exit on error in the case of unsupported keys
c23a34d - zeroize der(s) after try_from logic in rsa.rs in case of failure during the importing
956fe83 - zeroize key material in stored in the stack kx modules and not only in the heap
ddec026 - remove duplicate entry of RSA_PKCS1_SHA512 from the ALGORITHMS slice
26489a3 - make error clearer when expanding a slice in expand_block in hkdf.rs
91a0ead - removed manual free that was skipped on panic, using again the automatic drop calls via src/types/mod.rs
4390d38 - make hmacobject own its cleanup via the define_foreign_type_no_copy! macro in src/types/mod.rs
a05ad21 - free allocated Hmac object in case of failure from wc_HmacSetKey
e4878fe - removed sha*hmac modules, since they were merged into src/hmac/mod.rs which implements both
e759e19 - check returned value of wc_SignatureGetSize should be greater than zero on success (indicates maximum size of the signature)
f62dbb9 - fix data race in quic aes/chacha ciphers by allocating a fresh wolfcrypt context per encrypt/decrypt call instead of mutating shared C state
through &self
b4ae459 - remove derive(copy) from the chachaobject since no code copies it as of now
6172aec - fix heap leak of boxed aes/chacha in quic ciphers by freeing the box in aescipher/chachacipher drop and on the aesinit error path
54c0bf3 - reuse a single raw pointer for the in/out buffer in all aead encrypt/decrypt ffi calls to avoid stacked borrows provenance invalidation (UB under
miri)
088672d - added missing checks to validate sample length (quic) before encrypting both for aes and chacha20
8a20ba3 - run the quic tests in the ci/cd
6075ea9 - added missing checks in quic.rs against the tag_len, to make sure the cipher text is not malformed
427f921 - added proper checks against 24 bytes cipher text (aes128/256gcm)