[AMORO-2091] 0.4.x resolve the optimzing exception for KeyedTable with Timestamp(without zone) column as Primary key

[wangtaohz]

Why are the changes needed?

Close #2091 . fix in 0.4.x

Brief change log

• NodeFilter should use InternalRecordWrapper to build PrimaryKeyData

How was this patch tested?

• Add some test cases that check the changes thoroughly including negative and positive cases if possible

• Add screenshots for manual tests if appropriate

• Run test locally before making a pull request

Documentation

• Does this pull request introduce a new feature? (yes / no)
• If yes, how is the feature documented? (not applicable / docs / JavaDocs / not documented)

[codecov]

Codecov Report

Attention: 2 lines in your changes are missing coverage. Please review.

Files	Coverage Δ
...m/netease/arctic/io/reader/ArcticDeleteFilter.java	76.26% <0.00%> (-0.39%)	⬇️

📢 Thoughts on this report? Let us know!.

[zhoujinsong]

@wangtaohz Should we merge the changes to master branch too?

BTW，If the issue exist in master branch, we may fix it and merge back to 0.4.x.

[wangtaohz]

@wangtaohz Should we merge the changes to master branch too?
BTW，If the issue exist in master branch, we may fix it and merge back to 0.4.x.

Yes, the master has the same code, we'd better fix in master and cherry-pick to 0.4.x and also 0.5.x.

However, we can't reproduce it in master and 0.5.x (Mixed Format KeyedTable, Timestamp witout timezone as Primary Key: optimizing), because the sourceNodes for reader is null after version 0.5.x, so the NodeFilter didn't work in optimizing.

I also tried to reproduce it with MOR in flink and spark, and they all works well.

[wangtaohz]

We should not merge this PR, and I will cherry-pick #2095 to 0.4.x instead, then close this PR.