Replace ToCanonical trait usages with execute

[robert3005]

This also removes ArrayBuilder::extend_from_array and moves that logic
in to ArrayRef::append_to_builder

Plumb ExecutionCtx through more methods instead of creating fresh one.

Fixes: #3235

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

[codspeed-hq]

Merging this PR will degrade performance by 12.31%

⚠️ 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.

⚡ 2 improved benchmarks
❌ 9 regressed benchmarks
✅ 1584 untouched benchmarks
⏩ 4 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	extend_from_array_zctl[(1000, 8)]	241.1 µs	400.8 µs	-39.84%
❌	Simulation	extend_from_array_non_zctl_overlapping[(1000, 8)]	270 µs	435.6 µs	-38.02%
❌	Simulation	extend_from_array_non_zctl_overlapping[(1000, 32)]	731.8 µs	886.1 µs	-17.41%
❌	Simulation	slice_empty_vortex	339.4 ns	397.8 ns	-14.66%
❌	Simulation	extend_from_array_zctl[(1000, 64)]	1 ms	1.2 ms	-13.42%
❌	Simulation	i32_small_overlapping	39.8 µs	45 µs	-11.63%
❌	Simulation	extend_from_array_zctl[(10000, 8)]	1.9 ms	2.2 ms	-10.44%
❌	Simulation	extend_from_array_non_zctl_overlapping[(10000, 8)]	2.2 ms	2.5 ms	-10.29%
❌	Simulation	compact_sliced[(4096, 90)]	779.2 ns	867.2 ns	-10.15%
⚡	Simulation	bench_many_codes_few_values[1024]	528.1 µs	364.1 µs	+45.02%
⚡	Simulation	bitwise_not_vortex_buffer_mut[128]	273.6 ns	244.4 ns	+11.93%

Tip

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

---

_{Comparing claude/nice-hypatia-g68zgt (e582b35) with develop (733ab9e)}

Footnotes

1. 4 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. ↩

[lwwmanning]

builder_with_capacity(DType::List, ...) returns a ListViewBuilder<u64, u64>, so generic callers that allocate the canonical builder and then append a ListArray now hit this branch and error out instead of appending. For example, zip/case_when build the default list builder and append sliced branches through append_to_builder. This should either append into the canonical ListViewBuilder or fall back through execute::<ListViewArray> before appending.

[robert3005]

This part needs reworking. Needs to do a different downcast here

[lwwmanning]

len can now be as high as 2048, but random_strictly_sorted_ends still randomly chooses PType::U8 and later does u8::try_from(e).vortex_expect(...) after forcing the last end toward this target. Ordinary fuzzer input can therefore panic during Arbitrary instead of returning arbitrary::Error. Please choose an end type that can represent the generated max end, cap the target by the chosen type, or return an error instead of vortex_expect.