fix(llmrails): move LLM isolation setup to after KB initialization

[Pouyanpi]

Description

The LLM isolation code was called too early in __init__, before all components were fully initialized. This caused flow matching to fail when trying to resolve rail flow IDs. This issue was introduce in #1342

Move _create_isolated_llms_for_actions() call to after KB setup to ensure all initialization is complete before creating isolated LLMs.

[Copilot]

Pull Request Overview

Fixes a timing issue where LLM isolation setup was called too early in initialization, causing flow matching to fail when trying to resolve rail flow IDs. The fix moves the LLM isolation call to after knowledge base initialization to ensure all components are properly set up.

• Moved _create_isolated_llms_for_actions() call from _init_llms() to after KB setup in __init__()
• Added error handling for flow matching failures with fallback to all actions requiring LLMs
• Updated tests to remove prefix matching functionality and test the timing fix scenario

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
nemoguardrails/rails/llm/llmrails.py	Moves LLM isolation setup and adds error handling for flow matching
nemoguardrails/rails/llm/utils.py	Removes prefix matching logic from flow ID resolution
tests/test_rails_llm_utils.py	Updates tests to remove prefix matching test cases
tests/test_llm_isolation.py	Adds test for timing issue with empty flows scenario

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 71.59%. Comparing base (d1b35c8) to head (494dae3).
⚠️ Report is 1 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1348   +/-   ##
========================================
  Coverage    71.59%   71.59%           
========================================
  Files          168      168           
  Lines        16861    16862    +1     
========================================
+ Hits         12071    12072    +1     
  Misses        4790     4790           

Flag	Coverage Δ
python	71.59% <100.00%> (+<0.01%)	⬆️

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

Files with missing lines	Coverage Δ
nemoguardrails/rails/llm/llmrails.py	90.73% <100.00%> (+0.09%)	⬆️
nemoguardrails/rails/llm/utils.py	100.00% <100.00%> (ø)

🚀 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.

[tgasser-nv]

I added a couple of questions, could you take a look before merging? Also could you run a local test with a config that gave the error before the fix, and make sure it's fixed after this PR will be merged?

[Pouyanpi]

Also could you run a local test with a config that gave the error before the fix, and make sure it's fixed after this PR will be merged?

on develop branch:

from nemoguardrails import LLMRails, RailsConfig

config = RailsConfig.from_path("./examples/configs/nemoguards")

rails = LLMRails(config)

/Library/Caches/pypoetry/virtualenvs/nemoguardrails-BZdXtzBq-py3.12/lib/python3.12/site-packages/langchain_nvidia_ai_endpoints/_common.py:212: UserWarning: Found nvidia/llama-3.1-nemoguard-8b-content-safety in available_models, but type is unknown and inference may fail.
  warnings.warn(
Library/Caches/pypoetry/virtualenvs/nemoguardrails-BZdXtzBq-py3.12/lib/python3.12/site-packages/langchain_nvidia_ai_endpoints/_common.py:212: UserWarning: Found nvidia/llama-3.1-nemoguard-8b-topic-control in available_models, but type is unknown and inference may fail.
  warnings.warn(
Failed to create isolated LLMs for actions: No action found for flow_id: content safety check input $model=content_safety

after the fix:

from nemoguardrails import LLMRails, RailsConfig

config = RailsConfig.from_path("./examples/configs/nemoguards")

rails = LLMRails(config)

/Library/Caches/pypoetry/virtualenvs/nemoguardrails-BZdXtzBq-py3.12/lib/python3.12/site-packages/langchain_nvidia_ai_endpoints/_common.py:212: UserWarning: Found nvidia/llama-3.1-nemoguard-8b-content-safety in available_models, but type is unknown and inference may fail.
  warnings.warn(
/Library/Caches/pypoetry/virtualenvs/nemoguardrails-BZdXtzBq-py3.12/lib/python3.12/site-packages/langchain_nvidia_ai_endpoints/_common.py:212: UserWarning: Found nvidia/llama-3.1-nemoguard-8b-topic-control in available_models, but type is unknown and inference may fail.
  warnings.warn(