feat: add "stark-debug" feature

[jonathanpwang]

Adds a new "stark-debug" feature to openvm-circuit and openvm-sdk for debugging purposes: when turned on, we pay a performance penalty to run the stark-backend debugger on all trace matrices before generating the proof. The debugger checks the constraints logically instead of cryptographically, which gives more information failure information.

Also updated stark-backend to tag v1.2.0-rc.1 which updates to make logs at info level a little less verbose.

[github-actions]

group	app.proof_time_ms	app.cycles	app.cells_used	leaf.proof_time_ms	leaf.cycles	leaf.cells_used
verify_fibair	(-47 [-4.2%]) 1,084	322,637	17,671,652	-	-	-
fibonacci	(-45 [-2.1%]) 2,101	3,000,420	51,503,521	-	-	-
regex	(+55 [+0.8%]) 7,033	8,217,194	163,669,874	-	-	-
ecrecover	(-4 [-0.4%]) 1,125	274,568	8,596,011	-	-	-
pairing	(-36 [-1.0%]) 3,618	3,725,930	93,105,105	-	-	-

Commit: cc1fd38

Benchmark Workflow

[codspeed-hq]

CodSpeed WallTime Performance Report

Merging #1915 will degrade performances by 29.2%

_{Comparing feat/stark-debug (cc1fd38) with feat/new-execution (7cc545a)}

⚠️ 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.

Summary

⚡ 1 improvements
❌ 3 regressions
✅ 20 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
❌	benchmark_execute[bubblesort]	36.5 ms	41.5 ms	-12.18%
❌	benchmark_execute[fibonacci_iterative]	39.9 ms	52.6 ms	-24.11%
⚡	benchmark_execute[fibonacci_recursive]	80.5 ms	68.4 ms	+17.56%
❌	benchmark_execute[quicksort]	38.5 ms	54.4 ms	-29.2%