fix: collect_columns quadratic complexity

[crepererum]

Which issue does this PR close?

-

Rationale for this change

Fix accidental $O(n^2)$ in collect_columns.

What changes are included in this PR?

There are the following ways to insert a clone into a hash set:

• clone before check: Basically set.insert(x.clone()). That's rather expensive if you have a lot of duplicates.
• iter & clone: That's what we do right now, but that's in $O(n^2)$.
• check & insert: Basically if !set.contains(x) {set.insert(x.clone())} That requires two hash probes though.
• entry-based API: Sadly the stdlib doesn't really offer any handle/entry-based APIs yet (see Tracking issue for HashSet entry APIs rust-lang/rust#60896 ), but hashbrown does, so we can use the nice set.get_or_insert_owned(x) which will only clone the reference if it doesn't exists yet and only hashes once.

We now use the last approach.

Are these changes tested?

Functional tests still pass, performance change is only visible for large inputs (like 100+ columns).

Are there any user-facing changes?

⚠️ Breaking change: collect_columns now returns hashbrown::HashSet instead of std::collections::HashSet!

[alamb]

Looks good to me -- thank you @crepererum

I'll also run the planning benchmarks to see if I can see any improvement

[alamb]

I ran the benchmarks and don't see any negative (or positive) direct effect

Details

crepererum/fix_collect_columns_o2

Benchmarking logical_select_one_from_700: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 5.5s, enable flat sampling, or reduce sample count to 60.
logical_select_one_from_700
                        time:   [1.0805 ms 1.0827 ms 1.0849 ms]
Found 2 outliers among 100 measurements (2.00%)
  2 (2.00%) high mild

physical_select_one_from_700
                        time:   [3.4578 ms 3.4601 ms 3.4625 ms]
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) low mild
  3 (3.00%) high mild

logical_select_all_from_1000
                        time:   [17.965 ms 17.977 ms 17.990 ms]
Found 6 outliers among 100 measurements (6.00%)
  5 (5.00%) high mild
  1 (1.00%) high severe

physical_select_all_from_1000
                        time:   [43.469 ms 43.504 ms 43.547 ms]
Found 4 outliers among 100 measurements (4.00%)
  2 (2.00%) high mild
  2 (2.00%) high severe

Then main

Benchmarking logical_select_one_from_700: Warming up for 3.0000 s
Warning: Unable to complete 100 samples in 5.0s. You may wish to increase target time to 5.5s, enable flat sampling, or reduce sample count to 60.
logical_select_one_from_700
                        time:   [1.0891 ms 1.0908 ms 1.0926 ms]
Found 6 outliers among 100 measurements (6.00%)
  1 (1.00%) low severe
  2 (2.00%) low mild
  3 (3.00%) high mild

physical_select_one_from_700
                        time:   [3.5458 ms 3.5485 ms 3.5515 ms]
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) high mild
  1 (1.00%) high severe

logical_select_all_from_1000
                        time:   [18.000 ms 18.012 ms 18.025 ms]
Found 3 outliers among 100 measurements (3.00%)
  1 (1.00%) high mild
  2 (2.00%) high severe

physical_select_all_from_1000
                        time:   [43.609 ms 43.639 ms 43.671 ms]
Found 9 outliers among 100 measurements (9.00%)
  4 (4.00%) high mild
  5 (5.00%) high severe