ddl: refine pre-split region logic

[crazycs520]

What problem does this PR solve?

The original logic of pre_split_region will split 2^(pre_split_region -1) regions. It was strange for the user to understand and remember it.

This PR will simply split 2^(pre_split_region) regions when create table with pre_split_region.

What is changed and how it works?

Check List

Tests

• Unit test

Code changes

• Has exported function/method change

Side effects

Related changes

• Need to cherry-pick to the release branch

Release Note

• Change pre_split_regions logic for easy to understand and remember.

[codecov]

Codecov Report

Merging #11794 into master will not change coverage.
The diff coverage is n/a.

@@             Coverage Diff             @@
##             master     #11794   +/-   ##
===========================================
  Coverage   81.4701%   81.4701%           
===========================================
  Files           435        435           
  Lines         93719      93719           
===========================================
  Hits          76353      76353           
  Misses        11891      11891           
  Partials       5475       5475

[winkyao]

LGTM

[zimulala]

LGTM

[bb7133]

LGTM

[sre-bot]

/run-all-tests

[crazycs520]

Related docs PR: pingcap/docs-cn#1780

[sre-bot]

@crazycs520 merge failed.

[crazycs520]

/rebuild

[sre-bot]

cherry pick to release-3.0 in PR #11796

[sre-bot]

cherry pick to release-2.1 in PR #11797

[zz-jason]

@crazycs520 Can we make pre_split_region to be a boolean value? Let shard_row_id_bits to control the number of regions/shards to be split if that option is turned on?

[crazycs520]

@zz-jason For the compatibility, not suggest making pre_split_region to be a boolean value.
We plan to add 2 new option: row_id_shard_count , pre_split_region_count, then gradually deprecated pre_split_region and shard_row_id_bit