fix(update_replication_factor#12): update the mutation_2pc_min_replica_count base the max count of table

[foreverneverer]

Ref-Issue

#865

What problem does this PR solve?

#865 support different replica factor, but mutation_2pc_min_replica_count still keep 2, if one table update the factor to 2, when one replica is down(for example, offline node case), the table would keep write reject status.

So mutation_2pc_min_replica_count should be updated base the table factor, this pr should set it replica_max_count /2 + 1, but consider that sometimes user need set the replica_max_count > 3, I think the config mutation_2pc_min_replica_count is still valid:

1. if user set the mutation_2pc_min_replica_count > 0, the final 2pc_min_replica_count = mutation_2pc_min_replica_count
2. otherwise, 2pc_min_replica_count = app_max_replica_count <= 2 ? 1 : app_max_replica_count / 2 + 1 which can be make sure the table is alive though more node crush

Config-Change

if you want use value of config.ini, you don't need update it, otherwise:

- mutation_2pc_min_replica_count = 2
+ mutation_2pc_min_replica_count = 0

What is changed and how does it work?

Checklist

Tests

• Unit test
• Manual test (add detailed scripts or steps below)

1. start server in test cluster
4. create table with 3 replica count
5. start read/set with 10k qps
6. set replica count = 2
7. kill one node
8. watch the write status

Related changes

• Need to cherry-pick to the release branch
• Need to update the documentation
• Need to be included in the release note