pkcs5: return an error instead of panicking on a short DES/3DES IV

[arpitjain099]

While reading the PBES2 EncryptionScheme parser I noticed the DES-CBC and 3DES-EDE3-CBC arms can panic on attacker-controlled input.

Both DES arms do iv[0..DES_BLOCK_SIZE].try_into(), so the fixed-size slice happens before the try_into length check. If an AlgorithmIdentifier carries a DES-CBC or 3DES OID with an IV octet string shorter than 8 bytes, that range slice panics (range end index 8 out of range for slice of length N) instead of returning an error. The AES arms already call try_into on the whole slice, which reports a wrong length as an error, so this is just an inconsistency between the branches.

The IV comes straight from the DER parameters, so any code path that decodes an untrusted EncryptionScheme is affected, including pkcs8::EncryptedPrivateKeyInfo::decrypt, where the params are decoded before the password is ever used. A crafted 3DES-encrypted key blob panics regardless of the password.

The fix drops the manual slice and lets try_into on the full slice do the length check, matching the AES arms. DesCbc/DesEde3Cbc both hold [u8; 8], so TryInto<[u8; 8]> enforces exactly 8 bytes.

Added regression tests in pkcs5/tests/pbes2.rs: a short (3-byte) IV for both DES-CBC and 3DES now decodes to Err, and a valid 8-byte 3DES IV still decodes. The 3DES test is behind the 3des feature and the DES-CBC one behind des-insecure; ran with cargo test -p pkcs5 --features 3des,des-insecure. Before the change the short-IV cases panicked at pbes2.rs:546 and :552; after it they pass.