ref: split ast parsing / class exec loops

[jordanrfrazier]

Uses Artemis by TurinTech to find optimizations and quality improvements. These changes were selected as part of a Sourcekon profiling, after which Artemis Intelligence was used to suggest multiple versions of improvements based on iterative scoring feedback.

Specifically, these changes

• Utilize loop fission to improve readability, minimize operations in instruction cache to potentially improve cache locality, and pre-group definitions for a single compilation and execution.
• Improve error handling

[ogabrielluiz]

This looks good, @jordanrfrazier !

[jordanrfrazier]

@hughEvansOne ptal. Great work on Artemis. Looking forward to building better benchmarks so we can quantitatively see the improvements your team has on this project!

[codspeed-hq]

CodSpeed Performance Report

Merging #7248 will improve performances by 57.46%

_{Comparing various-qol-code-improvements (b05dfea) with main (0eeb142)}

Summary

⚡ 4 improvements
✅ 15 untouched benchmarks

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
⚡	test_build_flow_invalid_job_id	9 ms	7.9 ms	+13.97%
⚡	test_cancel_build_unexpected_error	899.9 ms	756.2 ms	+19%
⚡	test_cancel_build_with_cancelled_error	416.5 ms	264.5 ms	+57.46%
⚡	test_cancel_nonexistent_build	12 ms	8.2 ms	+46.37%