Tracking: Dedicated Batch Serving Cluster

[zwang28]

RFC: https://github.com/risingwavelabs/rfcs/blob/eric/dedicated-serving-cluster/rfcs/0014-dedicated-serving-cluster.md

We'll disable barrier read and only support reading from checkpoint. We ensure freshness by increasing checkpoint frequency.
We don't adopt approach 1 or approach 3, because we want different SLAs for streaming/serving cluster.

• use serving cluster in frontend feat(batch): support dedicated serving cluster #9040
• batch benchmark: read-write workload.
• improvement: better vnode mapping reorganization refactor(batch): maintain serving vnode mapping in meta node #10004
• improvement: availability feat(frontend): support mask failed serving worker temporarily #10328