Enable crash dump of input image

[grzegorz-roboflow]

What does this PR do?

When inference pipeline crashes it would be nice to be able to collect image/frame that caused the crash

Type of Change

• New feature (non-breaking change that adds functionality)

Testing

• I have tested this change locally

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code where necessary, particularly in hard-to-understand areas
• My changes generate no new warnings or errors
• I have updated the documentation accordingly (if applicable)

Additional Context

---

Note

Adds optional crash diagnostics for workflow execution.

• Introduces INFERENCE_DEBUG_OUTPUT_DIR env var in env.py to control where crash artifacts are written
• Implements _store_crash_info to write a .jpg of the input image and a .txt with exception + traceback
• Wraps SIMD batch step execution to capture exceptions and persist crash info when failures occur

^{Written by Cursor Bugbot for commit 1c62163. This will update automatically on new commits. Configure here.}

[grzegorz-roboflow]

bugbot run

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}