Rework VAR_LIST storage layout

[hououou]

Previously, we only stored the end offset in the VarList storage layout.
For an example layout for four lists of INT64: [4,7,8,12], null, [2, 3], []. We store as follows:

Offset column: [4, 4, 6, 6]
data column: [4, 7, 8, 12, 2, 3]

We need to ensure that the offset column must be in ascending order. So when we update [4,7,8,12] to [4,5,6,7,8], we need to rewrite the whole varlist column chunk, both offset and data column chunk.

Offset column: [5, 5, 7, 7]
data column: [4, 5, 6, 7, 8, 2, 3]

It will take time to rewrite the column chunk. To solve it, we additionally store the length of VarList. So the data layout is just as follows.

Offset column: [4, 4, 6, 6]
Size column: [4, 0, 2, 0]
data column: [4, 7, 8, 12, 2, 3]

When we update, we just append data in the back of the data column chunk and only update the offset and size column chunk. The varlist column chunk will be

Offset column: [10, 4, 6, 6]
Size column: [5, 0, 2, 0]
data column: [4, 7, 8, 12, 2, 3, 4, 5, 6, 7, 8]

[codecov]

Codecov Report

Attention: Patch coverage is 86.89320% with 27 lines in your changes are missing coverage. Please review.

Project coverage is 92.63%. Comparing base (c3decc2) to head (e4571a8).
Report is 4 commits behind head on master.

Files	Patch %	Lines
src/storage/store/var_list_column_chunk.cpp	72.36%	21 Missing ⚠️
src/storage/store/var_list_column.cpp	94.28%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3060      +/-   ##
==========================================
- Coverage   92.66%   92.63%   -0.04%     
==========================================
  Files        1157     1157              
  Lines       43074    43182     +108     
==========================================
+ Hits        39916    40003      +87     
- Misses       3158     3179      +21     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ray6080]

Why do you need a temp column chunk as a class field? Should remove it.

[ray6080]

Can you put some benchmark numbers here?

1. COPY node/rel with a var list property. You can populate some csv files of 10M tuples.
2. See if you can also benchmark inserting 1000 of nodes/rels after COPY.

[ray6080]

See #3093.