add region bucket informations in split_check.

[tonyxuqqi]

1. split_check.rs: add region_bucket information that are used for split large range query.
2. support both scan or approximate policy.

What problem does this PR solve?

improve large range query performance (such as select count(*), analyze table)
Issue Number: close #xxx

Problem Summary:

What is changed and how it works?

This is TiKV part change and all it does is basically update PD the current region buckets based on scan or approximate size. We virtually split the big region into region buckets, each region bucket is about 100MB size (the value is configurable).

Related changes

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