feat(mito): implements row group level page cache

[evenyag]

I hereby agree to the terms of the GreptimeDB CLA

What's changed and what's your intention?

This PR implements a row group-level page cache for the mito engine.

It adds a new page reader CachedPageReader that returns pages of a row group from the cache.

The first time we read a row group, we load all pages of the row group and put it into the cache. The next time we can fetch cached pages from the cache and build a CachedPageReader.

We use region id, file id, row group index, column index as cache key, which means we can only cache a column's row group level pages.

Cached pages are decompressed, so we can skip the decompression step and reduce 20% ~ 30% of the total scan time if a query hits the cache.

Checklist

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.

Refer to a related PR or issue link (optional)

[codecov]

Codecov Report

Merging #2688 (8d43cb7) into develop (730a3fa) will decrease coverage by 0.33%.
The diff coverage is 89.39%.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #2688      +/-   ##
===========================================
- Coverage    85.47%   85.14%   -0.33%     
===========================================
  Files          777      779       +2     
  Lines       125919   126339     +420     
===========================================
- Hits        107626   107573      -53     
- Misses       18293    18766     +473     

[fengjiachun]

LGTM

[killme2008]

LGTM. There is a confict @evenyag