feat: support to reject write after flushing (#4759)

* refactor: use `RegionRoleState` instead of `RegionState`

* feat: introducing `RegionLeaderState::Downgrading`

* refactor: introduce `set_region_role_state_gracefully`

* refactor: use `set_region_role` instead of `set_writable`

* feat: support to reject write after flushing

* fix: fix unit tests

* test: add unit test for `should_reject_write`

* chore: add comments

* chore: refine comments

* fix: fix unit test

* test: enable fuzz tests for Local WAL

* chore: add logs

* chore: rename `RegionStatus` to `RegionState`

* feat: introduce `DowngradingLeader`

* chore: rename

* refactor: refactor `set_role_state` tests

* test: ensure downgrading region will reject write

* chore: enhance logs

* chore: refine name

* chore: refine comment

* test: add tests for `set_role_role_state`

* fix: fix unit tests

* chore: apply suggestions from CR

* chore: apply suggestions from CR

.github/workflows/develop.yml

@@ -442,6 +442,13 @@ jobs:
             minio: true
             kafka: true
             values: "with-remote-wal.yaml"
+        include:
+          - target: "fuzz_migrate_mito_regions"
+            mode:
+              name: "Local WAL"
+              minio: true
+              kafka: false
+              values: "with-minio.yaml"
     steps:
       - name: Remove unused software
         run: |
@@ -530,7 +537,7 @@ jobs:
         with:
           image-registry: localhost:5001
           values-filename: ${{ matrix.mode.values }}
-          enable-region-failover: true
+          enable-region-failover: ${{ matrix.mode.kafka }}
       - name: Port forward (mysql)
         run: |
           kubectl port-forward service/my-greptimedb-frontend 4002:4002 -n my-greptimedb&