ListArray::reset_offsets doesn't need to revalidate array after reset

[robert3005]

Reset offsets operations if they start from a valid array they should stay valid

[codspeed-hq]

Merging this PR will improve performance by 20.02%

⚡ 16 improved benchmarks
✅ 1644 untouched benchmarks
⏩ 47 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	take_filter_list_random_mask_sequential_indices[768, 50]	132.8 µs	93.7 µs	+41.73%
⚡	Simulation	take_filter_list_random_mask_sequential_indices[256, 50]	134 µs	96 µs	+39.51%
⚡	Simulation	list_sum_small	144.5 µs	108 µs	+33.79%
⚡	Simulation	list_i32_large	212.7 µs	171.2 µs	+24.22%
⚡	Simulation	take_filter_list_slice_mask_sequential_indices[768, 50]	116.3 µs	93.8 µs	+24.03%
⚡	Simulation	take_filter_list_slice_mask_sequential_indices[256, 50]	115.4 µs	94 µs	+22.75%
⚡	Simulation	take_filter_list_slice_mask_random_indices[256, 50]	136.4 µs	118.6 µs	+15.03%
⚡	Simulation	take_filter_list_slice_mask_random_indices[768, 50]	136.7 µs	119.2 µs	+14.67%
⚡	Simulation	take_filter_list_random_mask_random_indices[256, 50]	140.1 µs	122.2 µs	+14.63%
⚡	Simulation	take_filter_list_random_mask_random_indices[768, 50]	144 µs	125.9 µs	+14.4%
⚡	Simulation	take_filter_list_small_random_mask_random_indices[768, 10]	137 µs	119.9 µs	+14.23%
⚡	Simulation	take_filter_list_small_uncached_random_mask_random_indices[256, 10]	133.5 µs	117 µs	+14.11%
⚡	Simulation	take_filter_list_small_random_mask_random_indices[256, 10]	132.6 µs	116.4 µs	+13.93%
⚡	Simulation	take_filter_list_small_uncached_random_mask_random_indices[768, 10]	136.8 µs	120.2 µs	+13.86%
⚡	Simulation	take_filter_list_nullable_random_mask_random_indices[256, 50]	145.6 µs	129.3 µs	+12.67%
⚡	Simulation	take_filter_list_nullable_random_mask_random_indices[768, 50]	150.5 µs	133.8 µs	+12.51%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing rk/resetunchecked (676c05b) with develop (f47bf54)}

Footnotes

1. 47 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. ↩