Tracking: make sure optd does not crash for TPC-H queries

[Gun9niR]

• Q1 feat: support IntervalMonthDayNano DF type #78
  • 🚧 Not working in perftest: Error: ArrowError(NotYetImplemented("Unsupported Interval Expression with value \"115 day (3) seconds\""))
• Q2 test: add tpc-h q2 #80
• Q3 test: add tpch 3, 15, 17 planner test #86
• Q4
  • optd panics with cannot find best binding for group (Not supporting left semi join)
• Q5 feat: support tpch q5 and q6 #69
• Q6 feat: support tpch q5 and q6 #69
  • 🚧 Working in optd but not in DataFusion: Error: Arrow error: Invalid argument error: Invalid comparison operation: Decimal128(15, 2) >= Decimal128(20, 0)
• Q7 feat: support tpch q7 and q9 #72
• Q8 test: add tpc-h q8 and fix expression parsing #64
• Q9 feat: support tpch q7 and q9 #72
• Q10 feat: support IntervalMonthDayNano DF type #78
• Q11
  • Non-equal join filer
• Q12 feat: support InListExpr #77
• Q13
  • DF Join filter says Filters applied during join (non-equi conditions), but there can actually be equal comparisons!
  • Non-equal join filter
• Q14 feat: support IntervalMonthDayNano DF type #78
• Q15 test: add tpch 3, 15, 17 planner test #86
  • 🚧 Not working in perftest: unsupported plan node: CreateView: Bare { table: "revenue0" }
• Q16
  • optd panics with cannot find best binding for group(Not supporting left anti join)
• Q17 test: add tpch 3, 15, 17 planner test #86
• Q18
  • optd panics with cannot find best binding for group (Not supporting left semi join)
• Q19 feat: support InListExpr #77
• Q20
  • optd panics with cannot find best binding for group (Not supporting left semi join)
• Q21
  • optd panics with cannot find best binding for group (Not supporting left semi and left anti join)
• Q22
  • optd panics with cannot find best binding for group(Not supporting left anti join)

Missing features

• Subquery: Q2, Q3, Q4, Q11, Q16, Q17, Q20, Q21, Q22
  • Correlated subquery is not supported in DF physical optimizer. DF's logical optimizer eliminates OuterReferenceColumn by moving the column to the outer query. Since we do not know how to access columns of the outer query, I think we should enable DF's logical optimizer when dealing with subqueries.
  • With DF's logical optimizer enabled, we get Error during planning: table 'datafusion.public.xxxx' not found error. When using explain_with_logical in the planner test, make sure to execute the DDLs with execute_with_logical!
• InListExpr: Q12, Q16, Q19 feat: support InListExpr #77
• IntervalMonthDayNano: Q1, Q10, Q14(feat: support IntervalMonthDayNano DF type #78)
• Multiple (non-equal) predicates for one join: Q11, Q13, Q22
• ColumnRefPropertyBulder
  • LogOp: Q16
• Left anti and left semi join: Q4, Q16, Q18, Q20, Q21, Q22

[jurplel]

We should test more TPC-H queries and make a list of what features we're missing.

[AveryQi115]

It seems q1 is failed for card test:

Error: Arrow error: Not yet implemented: Unsupported Interval Expression with value "115 day (3) seconds"

Caused by:
Not yet implemented: Unsupported Interval Expression with value "115 day (3) seconds"

[Gun9niR]

It seems q1 is failed for card test:

Error: Arrow error: Not yet implemented: Unsupported Interval Expression with value "115 day (3) seconds"

Caused by: Not yet implemented: Unsupported Interval Expression with value "115 day (3) seconds"

The error is thrown from SessionState::statement_to_plan, which means it's DF's problem, not optd's. We can modify the tpch template so it generates something DF supports.