Reduce number of clones and drops in expression optimisation logic

[robert3005]

Expression optimisation performs a lot of unnecessary clones and drops. Instead of having an extensible trait we limit the optimisation logic to only happen over Expressions and Arrays, logically these are the only two types we might want to run optimisations over. Using that simplification we convert majority of the allocations into borrows

[codspeed-hq]

Merging this PR will degrade performance by 1.1%

⚡ 4 improved benchmarks
❌ 6 regressed benchmarks
✅ 1952 untouched benchmarks
🆕 12 new benchmarks
⏩ 89 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	compact_sliced[(16384, 10)]	1.5 ms	2 ms	-27.28%
❌	Simulation	compact_sliced[(4096, 10)]	391.5 µs	528.5 µs	-25.92%
❌	Simulation	take[small_m/shuffled/primitive/nonnull/chunks=1024/indices=1000]	5.3 ms	6.1 ms	-13.22%
❌	Simulation	take[small_m/shuffled/primitive/nonnull/chunks=16384/indices=256]	3 ms	3.5 ms	-13.05%
❌	Simulation	take[small_m/shuffled/primitive/nonnull/chunks=1024/indices=256]	2 ms	2.3 ms	-12.53%
❌	Simulation	take[small_m/shuffled/primitive/nonnull/chunks=16384/indices=1000]	9 ms	10.2 ms	-12.19%
⚡	Simulation	optimize_or_chain[200]	1,177 µs	558.4 µs	×2.1
⚡	Simulation	chunked_dict_primitive_into_canonical[f32, (1000, 100, 100)]	1.8 ms	1.6 ms	+10.95%
⚡	Simulation	chunked_dict_primitive_into_canonical[f32, (1000, 10, 100)]	1.7 ms	1.5 ms	+10.87%
⚡	Simulation	take[small_m/shuffled/primitive/nonnull/chunks=16384/indices=16]	1.2 ms	1.1 ms	+10.63%
🆕	Simulation	optimize_lookup_predicate[ids=1, shape=balanced_or]	N/A	49.6 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=1, shape=in_list]	N/A	148.5 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=1024, shape=balanced_or]	N/A	2.3 ms	N/A
🆕	Simulation	optimize_lookup_predicate[ids=1024, shape=in_list]	N/A	50.5 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=16, shape=balanced_or]	N/A	93.4 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=16, shape=in_list]	N/A	48.5 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=256, shape=balanced_or]	N/A	616.6 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=256, shape=in_list]	N/A	50.4 µs	N/A
🆕	Simulation	optimize_lookup_predicate[ids=4096, shape=balanced_or]	N/A	8.9 ms	N/A
🆕	Simulation	optimize_lookup_predicate[ids=4096, shape=in_list]	N/A	57 µs	N/A
...	...	...	...	...	...

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

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing rk/exproptimise (fabbce3) with develop (a0a0b20)}

Footnotes

1. 89 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[robert3005]

we might need to change it again for layouts but this is at least the direction