fix: replace fragile text parsing with structured JSON output in performance workflow

[behrangsa]

Summary

This PR fixes the JSON parsing errors in the performance metrics workflow by replacing fragile text parsing with structured JSON output from cargo-criterion.

Root Cause

The performance workflow was failing with "Unexpected token m in JSON at position 172" because:

1. Fragile text parsing: Using grep/awk to extract benchmark values from text output
2. Empty values: When benchmarks don't run or fail, empty values create invalid JSON like "value": ,
3. No structured data: Regular cargo bench doesn't provide machine-readable output

Solution

🔧 Structured JSON Output

• Added cargo-criterion: Installs cargo-criterion tool that provides JSON output capability
• JSON message format: Uses --message-format=json to get structured benchmark data
• Proper parsing: Replaces grep/awk with jq for reliable JSON extraction

🛡️ Robust Error Handling

• Null value handling: Properly handles missing benchmark values with explicit null checks
• JSON validation: Ensures valid JSON is always generated even when benchmarks fail
• Fallback behavior: Gracefully handles cases where benchmark data is unavailable

Changes

Performance Workflow (.github/workflows/perf.yml)

• Install cargo-criterion: Added step to install cargo-criterion for JSON output
• JSON benchmark execution: Replace cargo bench with cargo criterion --message-format=json
• jq parsing: Replace fragile grep/awk with robust jq JSON extraction
• Safe JSON generation: Added proper null handling in performance-metrics.json creation

Key Improvements

• Reliable parsing: Uses structured JSON instead of text parsing
• Better error handling: Explicit handling of null/empty values
• Future-proof: Uses official JSON output format from cargo-criterion

Benefits

• ✅ Fixes JSON parsing errors that were causing CI failures
• ✅ More reliable metrics extraction using structured data
• ✅ Better error resilience with proper null handling
• ✅ Consistent data format across all benchmark runs
• ✅ Easier debugging with structured JSON output

Testing

Local testing confirms:

• JSON output is properly structured
• Benchmark values are correctly extracted (355.0168 ms from 355016848.38 ns)
• Null values are handled correctly
• Performance metrics JSON is valid

This addresses the failing build at: https://github.com/nutthead/samoid/actions/runs/16659795090/job/47153924965

[github-actions]

📊 Performance Test Report

Test Environment: Ubuntu Latest (GitHub Actions)
Commit: 381fc62
Branch: 34/merge
Triggered by: pull_request

📏 Binary Size Analysis (AC8.2)

Binary	Size	Status
samoid	1328832 bytes	✅
samoid-hook	989664 bytes	✅
Total	2318496 bytes	✅ < 10MB

🧠 Memory Usage Analysis (AC8.3)

Component	Memory Usage	Status
samoid init	4112 KB	✅
samoid-hook	2588 KB	✅
Limit	50 MB	✅ All under limit

⚡ Performance Benchmarks

Metric	Value	Target	Status
Hook Execution Overhead	.8906 ms	< 50ms	✅
Startup Time	TBD	< 100ms	✅
File Operations	TBD	Efficient	✅

📈 Performance Summary

• ✅ AC8.1: Hook execution overhead < 50ms
• ✅ AC8.2: Binary size < 10MB
• ✅ AC8.3: Memory usage < 50MB
• ✅ AC8.4: Startup time < 100ms
• ✅ AC8.5: Efficient file system operations

Full benchmark results available in workflow artifacts.

[github-actions]

🔒 Security Audit Report

❌ Error parsing audit report

Could not parse security audit results. Check the logs for details.

---

Security audit performed by cargo-audit

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!