refactor(stream): human-friendly StreamChunk literal for test

[wangrunji0408]

What's changed and what's your intention?

PLEASE DO NOT LEAVE THIS EMPTY !!!

Now it's painful to read unit tests of executors because the code to construct and verify StreamChunk is verbose.

This PR introduces a literal format for StreamChunk. An example is as follows:

   I I I   F      // a header with types: 'I' for i64, 'F' for f64
U- 2 5 .   .      // '.' for NULL
U+ 2 5 2 6.0 D    // 'D' for deleted in visibility
+  . . 4 8.0      // "//" for comments
-  . . 3 4.0

Developers can use StreamChunk::from_str to build a chunk from literal.
We also impl PartialEq for StreamChunk so that output chunks can be verified by assert_eq!.

Using these utilities, I rewrote a lot of code in the unit tests to make them more readable.

Checklist

• I have written necessary docs and comments
• I have added necessary unit tests and integration tests

Refer to a related PR or issue link (optional)

[CLAassistant]

All committers have signed the CLA.

[TennyZhuang]

I'd prefer to introduce a new trait StreamChunkTestExt, to avoid misuse.

Also, it's good to create a mod like test_prebuilt.

[TennyZhuang]

Rest LGTM

[codecov]

Codecov Report

Merging #2089 (e386989) into main (6019ef1) will decrease coverage by 0.39%.
The diff coverage is 99.57%.

@@            Coverage Diff             @@
##             main    #2089      +/-   ##
==========================================
- Coverage   71.03%   70.64%   -0.40%     
==========================================
  Files         635      635              
  Lines       81661    80464    -1197     
==========================================
- Hits        58010    56843    -1167     
+ Misses      23651    23621      -30     

Flag	Coverage Δ
rust	70.64% <99.57%> (-0.40%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/stream/src/executor_v2/chain.rs	76.08% <83.33%> (-2.22%)	⬇️
src/common/src/array/stream_chunk.rs	74.90% <94.93%> (-14.52%)	⬇️
src/common/src/array/column.rs	100.00% <100.00%> (ø)
src/common/src/array/data_chunk.rs	94.97% <100.00%> (ø)
src/stream/src/executor/dispatch.rs	77.89% <100.00%> (+0.77%)	⬆️
src/stream/src/executor/hash_join.rs	81.60% <100.00%> (-7.25%)	⬇️
src/stream/src/executor/source.rs	73.78% <100.00%> (-0.93%)	⬇️
src/stream/src/executor_v2/debug/schema_check.rs	94.33% <100.00%> (-0.40%)	⬇️
src/stream/src/executor_v2/debug/update_check.rs	93.75% <100.00%> (-0.28%)	⬇️
src/stream/src/executor_v2/filter.rs	89.82% <100.00%> (+0.49%)	⬆️
... and 30 more

📣 Codecov can now indicate which changes are the most critical in Pull Requests. Learn more

[BugenZhao]

Should we consider comparing two data chunks regardless of the visibility? That is, even though the visibilities differ, they are the same since the data after compaction is the same.

Seems this is used for test only.