Substrate storage query response does not seems to be correct for System.digests

Question

I am developing a series of Java tools for Substrate based chains, polkadot4j (it is currently under heavy development). I have problem in querying the storage for the pallet system and storage entry named "Digest". the storage key for this entry is 0x26aa394eea5630e07c48ae0c9558cef799e7f93fc6a98f0874fd057f111c4d2d, and the response I get for this key is: 0x080642414245b50103680100009d19291100000000a8691d03f5a419f00c91ce7d1b46385f9d86d3c6d67598560a0d1522245aba2f5590ae72b150586df25e56ef109a6fee3c656b4006512a52d0629a0901df9108a9b30875c15fc8370363616c376b1fecf9b164a299d50b2eceabbf5cc1d8c80d044245454684034dc81d0cb1830f54832f93ca94511a27fc6e8aafd2fc7cfcf93ed24393a1e4d2

this response should be decoded to Digest that is:

#[derive(PartialEq, Eq, Clone, Encode, Decode, RuntimeDebug, TypeInfo, Default)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub struct Digest {
    /// A list of logs in the digest.
    pub logs: Vec<DigestItem>,
}

and the DigestItem is :

#[derive(PartialEq, Eq, Clone, RuntimeDebug)]
pub enum DigestItem {
    /// A pre-runtime digest.
    ///
    /// These are messages from the consensus engine to the runtime, although
    /// the consensus engine can (and should) read them itself to avoid
    /// code and state duplication. It is erroneous for a runtime to produce
    /// these, but this is not (yet) checked.
    ///
    /// NOTE: the runtime is not allowed to panic or fail in an `on_initialize`
    /// call if an expected `PreRuntime` digest is not present. It is the
    /// responsibility of a external block verifier to check this. Runtime API calls
    /// will initialize the block without pre-runtime digests, so initialization
    /// cannot fail when they are missing.
    PreRuntime(ConsensusEngineId, Vec<u8>),

    /// A message from the runtime to the consensus engine. This should *never*
    /// be generated by the native code of any consensus engine, but this is not
    /// checked (yet).
    Consensus(ConsensusEngineId, Vec<u8>),

    /// Put a Seal on it. This is only used by native code, and is never seen
    /// by runtimes.
    Seal(ConsensusEngineId, Vec<u8>),

    /// Some other thing. Unsupported and experimental.
    Other(Vec<u8>),

    /// An indication for the light clients that the runtime execution
    /// environment is updated.
    ///
    /// Currently this is triggered when:
    /// 1. Runtime code blob is changed or
    /// 2. `heap_pages` value is changed.
    RuntimeEnvironmentUpdated,
}

so we excpect a sequence of DigestItem. now look again to the response, 0x080642414245b50103680100009d1929110000..., the first byte (08) is compact encoded of value 2, this is the sequence length, so we have 2 DigestItem, the second byte (06) represent the first DigestItem's variant (u8 encoded), that is the value 6, but DigestItem has 5 variants!

what is the problem? am I missing something?

note: https://polkadot.js.org/apps gives this result:

{
  logs: [
    {
      PreRuntime: [
        BABE
        0x03c90000004c1b291100000000549c0018aaa01e9516bb25a99404a2c3ead4d6f9b55f02f80aa4f78481e0a15e2742c8379c499c0816c7cee687bad5666d00634eaaa82435b62faac7eed7ec0d6daa797ef3469f08d345f9afa646706789cbe670eba23b96a9a938df82f35a08
      ]
    }
    {
      Consensus: [
        BEEF
        0x03d584f9d4530d898fcdfc83cb205654c66d10f011983b5bd7505c6a3637b829e7
      ]
    }
  ]
}

this is two items.

thank you