perf: avoid heap allocs reading Blake2s opcode operands

[orizi]

Read the Blake2s state and message via a new get_u32_array (const-generic stack array) instead of get_u32_range's Vec — two heap allocations per Blake instruction removed.

BREAKING: the Blake2sInvalidOperand error variant was unreachable (the vec always had the requested size on success) and is removed.

Based on #2391.

🤖 Generated with Claude Code

Performance

Expected: removes two heap allocations (a Vec<u32> of 8 and of 16) per Blake2s opcode instruction — relevant only to Stwo-style programs using the Blake opcode, roughly proportional to their Blake instruction density.

Measured (best-of-7 vs parent #2391 branch, local; noise ±2.4%): the standard benchmark suite contains no Blake-opcode programs, so the direct effect is not exercised. The suite nonetheless showed a consistent −1% to −9% (all 12 benchmarks negative, e.g. big_factorial −8.8%, big_fibonacci −7.3%); since the touched code is unreachable in these programs, that is most plausibly a code-layout/alignment side effect of the error-variant removal rather than an attributable algorithmic gain — treat the direct win as unmeasured pending a Blake-opcode workload.
Methodology: cairo_programs/benchmarks/*.cairo compiled with cairo-lang 0.14 --proof_mode, run via cairo-vm-cli <prog>.json --layout all_cairo --proof_mode, hyperfine best-of-7 per binary, local x86-64 Linux. Noise floor (identical-code control pair): ±2.4%.

---

This change is 

[github-actions]

Benchmark Results for unmodified programs 🚀

Command	Mean [s]	Min [s]	Max [s]	Relative
base big_factorial	1.957 ± 0.018	1.944	2.000	1.05 ± 0.02
head big_factorial	1.867 ± 0.027	1.849	1.941	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base big_fibonacci	1.859 ± 0.021	1.848	1.917	1.05 ± 0.01
head big_fibonacci	1.765 ± 0.010	1.750	1.778	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base blake2s_integration_benchmark	6.031 ± 0.236	5.857	6.611	1.05 ± 0.04
head blake2s_integration_benchmark	5.750 ± 0.085	5.682	5.973	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base compare_arrays_200000	1.921 ± 0.005	1.911	1.927	1.05 ± 0.01
head compare_arrays_200000	1.838 ± 0.009	1.827	1.850	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base dict_integration_benchmark	1.352 ± 0.005	1.345	1.359	1.03 ± 0.01
head dict_integration_benchmark	1.317 ± 0.018	1.301	1.363	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base field_arithmetic_get_square_benchmark	1.070 ± 0.002	1.067	1.075	1.02 ± 0.01
head field_arithmetic_get_square_benchmark	1.047 ± 0.005	1.041	1.058	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base integration_builtins	6.049 ± 0.086	6.001	6.288	1.04 ± 0.02
head integration_builtins	5.844 ± 0.035	5.806	5.892	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base keccak_integration_benchmark	6.047 ± 0.028	6.017	6.120	1.02 ± 0.01
head keccak_integration_benchmark	5.926 ± 0.021	5.897	5.950	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base linear_search	1.922 ± 0.012	1.914	1.953	1.04 ± 0.01
head linear_search	1.849 ± 0.011	1.833	1.875	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base math_cmp_and_pow_integration_benchmark	1.418 ± 0.005	1.414	1.429	1.03 ± 0.01
head math_cmp_and_pow_integration_benchmark	1.371 ± 0.006	1.364	1.382	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base math_integration_benchmark	1.390 ± 0.013	1.384	1.426	1.04 ± 0.01
head math_integration_benchmark	1.339 ± 0.007	1.329	1.348	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base memory_integration_benchmark	1.140 ± 0.002	1.137	1.143	1.04 ± 0.01
head memory_integration_benchmark	1.098 ± 0.006	1.089	1.107	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base operations_with_data_structures_benchmarks	1.443 ± 0.008	1.433	1.457	1.03 ± 0.01
head operations_with_data_structures_benchmarks	1.408 ± 0.009	1.395	1.425	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base pedersen	500.6 ± 1.2	498.8	502.9	1.01 ± 0.01
head pedersen	493.3 ± 2.3	491.2	497.5	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base poseidon_integration_benchmark	597.2 ± 2.2	593.1	599.7	1.03 ± 0.01
head poseidon_integration_benchmark	579.6 ± 3.6	575.6	589.1	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base secp_integration_benchmark	1.718 ± 0.019	1.707	1.769	1.02 ± 0.01
head secp_integration_benchmark	1.681 ± 0.010	1.670	1.704	1.00

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
base set_integration_benchmark	652.6 ± 1.1	650.8	653.9	1.02 ± 0.00
head set_integration_benchmark	638.4 ± 1.9	636.8	642.3	1.00

Command	Mean [s]	Min [s]	Max [s]	Relative
base uint256_integration_benchmark	3.621 ± 0.044	3.590	3.720	1.03 ± 0.01
head uint256_integration_benchmark	3.500 ± 0.027	3.477	3.563	1.00

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.23%. Comparing base (c610d41) to head (2c43ce0).

Additional details and impacted files

@@                           Coverage Diff                           @@
##           claude/vm-execution-performance-32hi0c    #2397   +/-   ##
=======================================================================
  Coverage                                   96.23%   96.23%           
=======================================================================
  Files                                         107      107           
  Lines                                       37985    37990    +5     
=======================================================================
+ Hits                                        36555    36560    +5     
  Misses                                       1430     1430           

☔ 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.