Improve intersect_by_rank performance

[robert3005]

We never spent time and this is useful for merging selections and filters in
scans

Signed-off-by: Robert Kruszewski github@robertk.io

[robert3005]

I will go over this tomorrow @joseph-isaacs I looked at #7098 and #7393 which both optimised slightly different cases of this function. I tried to combine the two.

[codspeed-hq]

Merging this PR will improve performance by ×19

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚡ 25 improved benchmarks
✅ 1181 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	WallTime	cuda/bitpacked_u8/unpack/3bw[100M]	352.3 µs	300.4 µs	+17.24%
⚡	Simulation	density_matrix[(0.05, 0.05, "self_sparse_mask_sparse")]	83 µs	47.6 µs	+74.43%
⚡	Simulation	density_matrix[(0.5, 0.05, "self_dense_mask_sparse")]	482.5 µs	53.2 µs	×9.1
⚡	Simulation	intersect_by_rank[(10000, "random")]	103.6 µs	10.3 µs	×10
⚡	Simulation	intersect_by_rank[(100000, "random")]	979.4 µs	53 µs	×18
⚡	Simulation	density_matrix[(0.05, 0.5, "self_sparse_mask_dense")]	131.8 µs	47.6 µs	×2.8
⚡	Simulation	density_matrix[(0.5, 0.5, "self_dense_mask_dense")]	979 µs	52.8 µs	×19
⚡	Simulation	intersect_by_rank[(10000, "runs")]	103.6 µs	10.1 µs	×10
⚡	Simulation	intersect_by_rank[(100000, "runs")]	976.8 µs	53 µs	×18
⚡	Simulation	rank_indices[(0.05, 0.05, "self_sparse_rank_sparse")]	80.9 µs	43 µs	+87.87%
⚡	Simulation	rank_indices[(0.5, 0.01, "self_dense_rank_very_sparse")]	427.9 µs	58.8 µs	×7.3
⚡	Simulation	rank_indices[(0.5, 0.5, "self_dense_rank_dense")]	867.5 µs	53.4 µs	×16
⚡	Simulation	sparse[(100000, 0.05, "sparse_5pct")]	132.1 µs	47.8 µs	×2.8
⚡	Simulation	sparse[(100000, 0.5, "dense_50pct")]	979.7 µs	53.2 µs	×18
⚡	Simulation	very_sparse_mask_cached[(0.5, 0.005, "self_dense_mask_0p5pct")]	422.3 µs	50.6 µs	×8.4
⚡	Simulation	very_sparse_mask_cached[(0.5, 0.02, "self_dense_mask_2pct")]	435.5 µs	73.7 µs	×5.9
⚡	Simulation	very_sparse_mask_uncached[(0.5, 0.005, "self_dense_mask_0p5pct")]	432.3 µs	59.5 µs	×7.3
⚡	Simulation	very_sparse_mask_uncached[(0.5, 0.02, "self_dense_mask_2pct")]	449 µs	82.1 µs	×5.5
⚡	Simulation	rank_indices[(0.05, 0.5, "self_sparse_rank_dense")]	120.1 µs	47.4 µs	×2.5
⚡	Simulation	rank_indices[(0.5, 0.05, "self_dense_rank_sparse")]	462.6 µs	58.6 µs	×7.9
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

---

_{Comparing rk/intersect-by-rank (a8488fc) with develop (f307edc)}