perf: cache compiled hint data on the runner

[orizi]

get_hint_data compiled every hint in the program on each run call; proof-mode trace padding calls run_for_steps/run_until_next_power_of_2 in a loop, recompiling all hints per padding step. Compile once on first use and reuse across the runner's run calls (assumes the same hint processor across a runner's runs).

Based on #2392.

🤖 Generated with Claude Code

---

This change is 

[github-actions]

Benchmark Results for unmodified programs 🚀

Command	Mean [s]	Min [s]	Max [s]	Relative
base big_factorial	1.883 ± 0.044	1.840	1.968	1.02 ± 0.03
head big_factorial	1.847 ± 0.031	1.815	1.919	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base big_fibonacci	1.784 ± 0.031	1.732	1.831	1.00 ± 0.03
head big_fibonacci	1.780 ± 0.037	1.743	1.845	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base blake2s_integration_benchmark	5.349 ± 0.073	5.254	5.462	1.00 ± 0.02
head blake2s_integration_benchmark	5.327 ± 0.086	5.233	5.538	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base compare_arrays_200000	1.866 ± 0.033	1.834	1.931	1.01 ± 0.02
head compare_arrays_200000	1.851 ± 0.027	1.813	1.898	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base dict_integration_benchmark	1.218 ± 0.011	1.202	1.239	1.00
head dict_integration_benchmark	1.233 ± 0.033	1.190	1.300	1.01 ± 0.03

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base field_arithmetic_get_square_benchmark	881.6 ± 15.7	869.6	920.9	1.00 ± 0.02
head field_arithmetic_get_square_benchmark	878.0 ± 13.4	865.5	899.9	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base integration_builtins	5.313 ± 0.167	5.166	5.751	1.00
head integration_builtins	5.326 ± 0.147	5.207	5.679	1.00 ± 0.04

Command	Mean [s]	Min [s]	Max [s]	Relative
base keccak_integration_benchmark	5.521 ± 0.091	5.403	5.677	1.00
head keccak_integration_benchmark	5.634 ± 0.194	5.364	6.120	1.02 ± 0.04

Command	Mean [s]	Min [s]	Max [s]	Relative
base linear_search	1.873 ± 0.027	1.846	1.939	1.00
head linear_search	1.894 ± 0.048	1.856	1.984	1.01 ± 0.03

Command	Mean [s]	Min [s]	Max [s]	Relative
base math_cmp_and_pow_integration_benchmark	1.385 ± 0.026	1.357	1.438	1.00 ± 0.02
head math_cmp_and_pow_integration_benchmark	1.385 ± 0.022	1.356	1.431	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base math_integration_benchmark	1.302 ± 0.036	1.267	1.373	1.00 ± 0.03
head math_integration_benchmark	1.297 ± 0.015	1.279	1.330	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base memory_integration_benchmark	1.128 ± 0.030	1.087	1.176	1.02 ± 0.03
head memory_integration_benchmark	1.102 ± 0.022	1.081	1.136	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base operations_with_data_structures_benchmarks	1.304 ± 0.021	1.274	1.338	1.00 ± 0.02
head operations_with_data_structures_benchmarks	1.304 ± 0.020	1.283	1.350	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base pedersen	395.7 ± 9.3	388.7	419.2	1.01 ± 0.03
head pedersen	389.9 ± 4.8	380.7	397.7	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base poseidon_integration_benchmark	511.2 ± 19.7	498.4	563.8	1.00
head poseidon_integration_benchmark	514.1 ± 10.6	504.6	531.2	1.01 ± 0.04

Command	Mean [s]	Min [s]	Max [s]	Relative
base secp_integration_benchmark	1.330 ± 0.023	1.310	1.377	1.00
head secp_integration_benchmark	1.343 ± 0.032	1.318	1.417	1.01 ± 0.03

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base set_integration_benchmark	513.4 ± 22.1	481.2	544.1	1.03 ± 0.04
head set_integration_benchmark	498.2 ± 3.1	494.0	502.8	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base uint256_integration_benchmark	3.335 ± 0.080	3.255	3.542	1.00 ± 0.03
head uint256_integration_benchmark	3.333 ± 0.035	3.263	3.385	1.00

[codecov]

Codecov Report

❌ Patch coverage is 83.33333% with 3 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (orizi/09-09-fix_run_hints_of_each_visited_pc_in_run_for_steps@681c028). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
vm/src/vm/runners/cairo_runner.rs	83.33%	3 Missing ⚠️

Additional details and impacted files

@@                                       Coverage Diff                                        @@
##             orizi/09-09-fix_run_hints_of_each_visited_pc_in_run_for_steps    #2393   +/-   ##
================================================================================================
  Coverage                                                                 ?   96.23%           
================================================================================================
  Files                                                                    ?      107           
  Lines                                                                    ?    38018           
  Branches                                                                 ?        0           
================================================================================================
  Hits                                                                     ?    36585           
  Misses                                                                   ?     1433           
  Partials                                                                 ?        0           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[orizi]

Closing in favor of a re-filed PR: this PR's cache scope has a stuck cache reservation (Unable to reserve cache with key cairo_bench_programs-..., left by a cancelled duplicate run) that blocks every bench-programs cache save, so merge-caches can never succeed here. Same branch, fresh PR = fresh cache scope.