WIP: Avoid copying LogicalPlans / Exprs during OptimizerPasses

[alamb]

Which issue does this PR close?

Related to #9637

Rationale for this change

@jayzhan211 @mustafasrepo and I have been discussing how to speed up planning by avoiding copying Exprs / LogicalPlans

What changes are included in this PR?

This PR is a prototype of a suggestion here #9637 (comment) to see if I can get a significant boost in Expr simplify

The theory is that if I can rerite the optimizer to take ownership of the plans they can be rewritten in place

If this works out, I'll turn this into an actual PR for review.

Update: this approach improves planning speed by 25%

Are these changes tested?

NA

Timings (main):

cargo bench   --bench sql_planner -- physical_plan_tpch_all

Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 7.4s, or reduce sample count to 60.
physical_plan_tpch_all  time:   [70.511 ms 71.102 ms 71.828 ms]
                        change: [-1.3186% +2.1648% +4.6865%] (p = 0.17 > 0.05)
                        No change in performance detected.
Found 9 outliers among 100 measurements (9.00%)
  4 (4.00%) low mild
  1 (1.00%) high mild
  4 (4.00%) high severe

This branch

Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 5.4s, or reduce sample count to 90.
physical_plan_tpch_all  time:   [54.371 ms 54.512 ms 54.675 ms]
                        change: [-24.152% -23.333% -22.648%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  3 (3.00%) high mild
  1 (1.00%) high severe

Are there any user-facing changes?

[alamb]

rewriting the optimizer passes to take in owned LogicalPlan is the key idea

I haven't plumbed it completely through yet, but the approach is looking promising

[alamb]

Update: I think this PR now works well enough to show this is a promising approach. Specifically, by just fixing SimplifyExprs to not copy plans around, the planning benchmark goes 25% faster. I am pretty sure if we fix common subexpr eliminate and a few other passes we could get the benchmark running 2x as fast

physical_plan_tpch_all  time:   [54.371 ms 54.512 ms 54.675 ms]
                        change: [-24.152% -23.333% -22.648%] (p = 0.00 < 0.05)
                        Performance has improved.

The code in this PR is atrocious, but I think works well enough to demonstrate the idea

The core problem is that when the LogicalPasses work, they copy the LogicalPlans many times (e.g. each call to new_with_exprs or new_with_children results in Copying the entire node and its embedded exprs (though not its children as they are Arcd)

cc @sadboy who I think has observed this before as well

[alamb]

Continuing in #9780, so closing this one