Add support for correlated subqueries & fix all related TPC-H benchmark issues

[avantgardnerio]

Which issue does this PR close?

Closes #159
Closes #160
Closes #163
Closes #168
Closes #171
Closes #172
Closes #175

Rationale for this change

Getting the TPC-H queries to work allows Datafusion users and potential users to compare standard benchmarks to decide if Datafusion is the correct choice for their use case.

What changes are included in this PR?

3 new query optimizers:

1. decorrelate_scalar_subquery
2. decorrelate_where_exists
3. decorrelate_where_in

pros

• all TPC-H queries pass but 15 (create view, multiple statements) and 21 (error in filter push down)
• it appears to be a generaliz(ed/able) approach to de-correlating subqueries
• it handles things like multiple subqueries per filter, recusive subqueries, correlated and uncorrelated, etc
• with the date stuff merged, it now largely works out of the box

cons

• there is insufficient unit testing - update: I've added a bunch of unit tests, but fuzzing may be required to merge something like this
• the code is verbose, and duplicated - update: it's definitely less verbose, and less duplicated, thoughts on the new methods?
• variables names are hard to understand
• it overlaps with filter_to_join

Are there any user-facing changes?

Correlated subqueries should work , but likely some will break or return incorrect results (thus the "draft" status).

[codecov-commenter]

Codecov Report

Merging #2885 (e341570) into master (b49093c) will increase coverage by 0.19%.
The diff coverage is 93.29%.

@@            Coverage Diff             @@
##           master    #2885      +/-   ##
==========================================
+ Coverage   85.43%   85.62%   +0.19%     
==========================================
  Files         275      279       +4     
  Lines       49739    50872    +1133     
==========================================
+ Hits        42495    43561    +1066     
- Misses       7244     7311      +67     

Impacted Files	Coverage Δ
...tafusion/core/src/physical_plan/file_format/mod.rs	97.36% <ø> (ø)
datafusion/optimizer/src/utils.rs	89.07% <89.28%> (+0.50%)	⬆️
datafusion/optimizer/src/test/mod.rs	94.23% <92.10%> (-5.77%)	⬇️
...usion/optimizer/src/decorrelate_scalar_subquery.rs	92.75% <92.75%> (ø)
...tafusion/optimizer/src/decorrelate_where_exists.rs	93.44% <93.44%> (ø)
datafusion/expr/src/logical_plan/plan.rs	77.43% <93.75%> (+0.10%)	⬆️
datafusion/core/tests/sql/mod.rs	98.26% <94.02%> (-0.56%)	⬇️
datafusion/optimizer/src/decorrelate_where_in.rs	94.13% <94.13%> (ø)
datafusion/core/tests/sql/subqueries.rs	94.32% <94.32%> (ø)
datafusion/common/src/error.rs	80.45% <100.00%> (+0.45%)	⬆️
... and 11 more

[avantgardnerio]

I think this is converging on a version I believe would be a good start on decorrelating sub-queries (it doesn't have to support all cases, as long as it fails instead of returning invalid plans). I've spent about 3 weeks on it so far, and I really would appreciate a thorough review before going too much farther. The longer I stare at it without external input, the more blind I will be as to what I missed or could do differently.

@andygrove @Dandandan @alamb I'd really appreciate your input (especially failing test case suggestions and readability improvements, but also thoughts on general approach, relationship with other work, and what should be in/out of scope).

[alamb]

Any last thoughts @Dandandan @andygrove ?

[Dandandan]

Thanks for the implementation @avantgardnerio awesome addition to DataFusion 😎

[avantgardnerio]

I think all the items listed as critical are now resolved.

[alamb]

Unless there are any other comments, I'll plan to merge this PR tomorrow

[alamb]

👍

[alamb]

Thanks again @avantgardnerio

[ursabot]

Benchmark runs are scheduled for baseline = 117df4d and contender = 7b0f2f8. 7b0f2f8 is a master commit associated with this PR. Results will be available as each benchmark for each run completes.
Conbench compare runs links:
[Skipped ⚠️ Benchmarking of arrow-datafusion-commits is not supported on ec2-t3-xlarge-us-east-2] ec2-t3-xlarge-us-east-2
[Skipped ⚠️ Benchmarking of arrow-datafusion-commits is not supported on test-mac-arm] test-mac-arm
[Skipped ⚠️ Benchmarking of arrow-datafusion-commits is not supported on ursa-i9-9960x] ursa-i9-9960x
[Skipped ⚠️ Benchmarking of arrow-datafusion-commits is not supported on ursa-thinkcentre-m75q] ursa-thinkcentre-m75q
Buildkite builds:
Supported benchmarks:
ec2-t3-xlarge-us-east-2: Supported benchmark langs: Python, R. Runs only benchmarks with cloud = True
test-mac-arm: Supported benchmark langs: C++, Python, R
ursa-i9-9960x: Supported benchmark langs: Python, R, JavaScript
ursa-thinkcentre-m75q: Supported benchmark langs: C++, Java