Reorder agg kernel dispatch, and have Combined use inner accumulators

[gatesn]

Aggregate kernel dispatch previously tried the vtable before the plugin registry, meaning plugins could not intercept the aggregate function as they can with array execution.

This change also fixed Combined to use inner accumulators, instead of just inner partials. This fixes the kernel dispatch for the delegated aggregate functions. e.g. a custom Sum<->Dict kernel previously would not have been used for Combiner<->Dict, even though it delegates to Sum + Count.

[codspeed-hq]

Merging this PR will improve performance by 31.26%

⚠️ 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.

⚡ 24 improved benchmarks
✅ 1186 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	take_search[(0.005, 0.05)]	168.3 µs	131.7 µs	+27.8%
⚡	Simulation	take_search[(0.005, 0.1)]	320.3 µs	247.2 µs	+29.53%
⚡	Simulation	take_search[(0.005, 0.5)]	1.5 ms	1.2 ms	+31.03%
⚡	Simulation	take_search[(0.005, 1.0)]	3.1 ms	2.3 ms	+31.26%
⚡	Simulation	take_search[(0.01, 0.05)]	179.4 µs	142.8 µs	+25.64%
⚡	Simulation	take_search[(0.01, 0.1)]	341.2 µs	268.1 µs	+27.23%
⚡	Simulation	take_search[(0.01, 0.5)]	1.6 ms	1.3 ms	+28.58%
⚡	Simulation	take_search[(0.01, 1.0)]	3.3 ms	2.5 ms	+28.77%
⚡	Simulation	take_search[(0.1, 0.05)]	249.1 µs	212.4 µs	+17.23%
⚡	Simulation	take_search[(0.1, 0.1)]	458.6 µs	385.6 µs	+18.95%
⚡	Simulation	take_search[(0.1, 0.5)]	2.2 ms	1.8 ms	+20.4%
⚡	Simulation	take_search[(0.1, 1.0)]	4.3 ms	3.5 ms	+20.63%
⚡	Simulation	take_search_chunked[(0.005, 0.05)]	193.1 µs	162.1 µs	+19.14%
⚡	Simulation	take_search_chunked[(0.005, 0.1)]	369.1 µs	307.2 µs	+20.17%
⚡	Simulation	take_search_chunked[(0.005, 0.5)]	1.8 ms	1.5 ms	+21.06%
⚡	Simulation	take_search_chunked[(0.005, 1.0)]	3.5 ms	2.9 ms	+21.19%
⚡	Simulation	take_search_chunked[(0.01, 0.05)]	206.2 µs	175.2 µs	+17.71%
⚡	Simulation	take_search_chunked[(0.01, 0.1)]	393.9 µs	331.9 µs	+18.67%
⚡	Simulation	take_search_chunked[(0.01, 0.5)]	1.9 ms	1.6 ms	+19.48%
⚡	Simulation	take_search_chunked[(0.01, 1.0)]	3.8 ms	3.2 ms	+19.59%
...	...	...	...	...	...

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

---

_{Comparing ngates/aggregate-kernels (842b03a) with develop (e160125)}