Bug Report

When slow logs exist in the processing of StoreFsm::handle_msgs, the relevant raftstore
will print the corresponding StoreMsg for the developer to confirm which is the slowest
StoreMsg.

However, the current implementation prints the StoreMsg with incorrect index, generating
misleading debugging log.

      // for `StoreFsm::handle_msgs`
      let mut distribution = [0; StoreMsg::<EK>::COUNT];
      for m in msgs.drain(..) {
            distribution[m.discriminant()] += 1;
            ...
      }
       slow_log!(
            T timer,
            "[store {}] handle {} store messages {:?}",
            self.fsm.store.id,
            count,
            StoreMsg::<EK>::VARIANTS.iter().zip(distribution).filter(|(_, c)| *c > 0).format(", "),
        );

The root cause is that the m.discriminant() mismatches with the definition of StoreMsg:

For example, the slow StoreMsg should be StoreMsg::CompactEvent rather than StoreMsg::Unreachable,
but the logs are:

[2025/06/15 18:16:38.328 +00:00] [WARN] [store.rs:902] ["[store 22] handle 1 store messages (\"StoreUnreachable\", 1)"] [takes=1330] [thread_id=204]

What version of TiKV are you using?

v7.5.6

What operating system and CPU are you using?

Centos 7

Steps to reproduce

What did you expect?

The slow log should be StoreMsg::CompactEvent rather than StoreMsg::Unreachable.

What did happened?

The slow log is StoreMsg::Unreachable.