feat(expr): Implement IS [NOT] DISTINCT FROM expression

[Graphcalibur]

What's changed and what's your intention?

• This PR implements the IS [NOT] DISTINCT FROM expression.
• This was done by using the existing implementations for expressions such as IsNull, IsNotNull, Equals, and NotEquals to build a binding with the same truth value as IS NOT DISTINCT FROM.

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)

#2513

[CLAassistant]

All committers have signed the CLA.

[codecov]

Codecov Report

Merging #2582 (8c1f9cf) into main (8d0d69b) will increase coverage by 0.17%.
The diff coverage is 30.76%.

@@            Coverage Diff             @@
##             main    #2582      +/-   ##
==========================================
+ Coverage   72.19%   72.37%   +0.17%     
==========================================
  Files         674      678       +4     
  Lines       87853    88038     +185     
==========================================
+ Hits        63422    63714     +292     
+ Misses      24431    24324     -107     

Flag	Coverage Δ
rust	72.37% <30.76%> (+0.17%)	⬆️

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

Impacted Files	Coverage Δ
src/frontend/src/binder/expr/mod.rs	73.20% <30.76%> (-12.28%)	⬇️
src/bench/ss_bench/utils/my_stats.rs	0.00% <0.00%> (-72.42%)	⬇️
src/batch/src/executor2/generate_series.rs	0.00% <0.00%> (-68.86%)	⬇️
src/batch/src/executor2/generate_time_series.rs	66.66% <0.00%> (-5.75%)	⬇️
src/common/src/types/chrono_wrapper.rs	72.91% <0.00%> (-1.35%)	⬇️
src/storage/src/monitor/state_store_metrics.rs	98.17% <0.00%> (-1.34%)	⬇️
src/storage/src/hummock/sstable_store.rs	65.11% <0.00%> (-0.81%)	⬇️
src/frontend/src/scheduler/execution/stage.rs	45.88% <0.00%> (-0.61%)	⬇️
src/expr/src/expr/build_expr_from_prost.rs	69.50% <0.00%> (-0.22%)	⬇️
... and 45 more

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

[lmatz]

LGTM

[lmatz]

Is there any preference for new_unchecked over new or the other way around? @xiangjinwu

[xiangjinwu]

As the name and doc suggested, new_unchecked bypasses type checking and should only be used for a good reason.
Here the innermost expressions (IsNull, IsNotNull, Equal, NotEqual) are using new correctly, and outer expressions (And, Or) are bypassing the checks because both operands are from validated inner expressions. This is acceptable, but maybe the performance gain is negligible.

[xiangjinwu]

Btw each clone of expr means repeated evaluation in the backend. It may be better to implement these 2 operators in backend directly rather than rewriting to such long expressions in the frontend.

[lmatz]

I see, let this PR add some tests first to make sure the semantics are all correct.

Then another PR implements the operators in the backend and removes the rewriting.

[Graphcalibur]

The expression binding is fixed so that it can now only evaluate to either true or false (instead of unknown). E2E tests were also added.