fix(jailbreak): pin numpy==1.23.5 for scikit-learn compatibility

[Pouyanpi]

Problem

The Docker build for the jailbreak detection heuristics server was failing with a binary incompatibility error:

ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

This error occurred during the step that pre downloads the GPT-2 model, when the transformers library tried to import scikit-learn, which had been compiled against a different version of numpy than what was present at runtime.

Root Cause

• The requirements.txt had scikit-learn==1.2.2 pinned but no explicit numpy version constraint
• during Docker build transformers and other deps pulled in a newer version of numpy
• scikit-learn 1.2.2 was compiled against numpy 1.23.x series, but a newer numpy version was installed
• this created a binary incompatibility between scikit-learn's C extensions and the runtime numpy version

[Pouyanpi]

related todo: #934

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 69.34%. Comparing base (c9cfa7e) to head (da6cfc7).

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1249   +/-   ##
========================================
  Coverage    69.34%   69.34%           
========================================
  Files          161      161           
  Lines        15995    15995           
========================================
  Hits         11091    11091           
  Misses        4904     4904           

Flag	Coverage Δ
python	69.34% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

🚀 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.

[erickgalinkin]

LGTM