Add comprehensive Ollama log checking and configurable LLM judge mode

Test case enhancements:
- TC-RUNTIME-001: Add startup log error checking (CUDA, CUBLAS, CPU fallback)
- TC-RUNTIME-002: Add GPU detection verification, CUDA init checks, error detection
- TC-RUNTIME-003: Add server listening verification, runtime error checks
- TC-INFERENCE-001: Add model loading logs, layer offload verification
- TC-INFERENCE-002: Add inference error checking (CUBLAS/CUDA errors)
- TC-INFERENCE-003: Add API request log verification, response time display

Workflow enhancements:
- Add judge_mode input (simple/llm/dual) to all workflows
- Add judge_model input to specify LLM model for judging
- Configurable via GitHub Actions UI without code changes

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

.github/workflows/runtime.yml

@@ -11,6 +11,20 @@ on:
         options:
           - 'true'
           - 'false'
+      judge_mode:
+        description: 'Test judge mode'
+        required: false
+        default: 'simple'
+        type: choice
+        options:
+          - 'simple'
+          - 'llm'
+          - 'dual'
+      judge_model:
+        description: 'LLM model for judging (if llm/dual mode)'
+        required: false
+        default: 'gemma3:4b'
+        type: string
   workflow_call:      # Called by other workflows
     inputs:
       keep_container:
@@ -18,6 +32,16 @@ on:
         required: false
         default: false
         type: boolean
+      judge_mode:
+        description: 'Test judge mode (simple, llm, dual)'
+        required: false
+        default: 'simple'
+        type: string
+      judge_model:
+        description: 'LLM model for judging'
+        required: false
+        default: 'gemma3:4b'
+        type: string
     outputs:
       result:
         description: "Runtime test result"
@@ -53,8 +77,23 @@ jobs:
         id: runtime-tests
         run: |
           cd tests
+
+          # Build judge flags based on input
+          JUDGE_FLAGS=""
+          if [ "${{ inputs.judge_mode }}" = "simple" ] || [ -z "${{ inputs.judge_mode }}" ]; then
+            JUDGE_FLAGS="--no-llm"
+          elif [ "${{ inputs.judge_mode }}" = "dual" ]; then
+            JUDGE_FLAGS="--dual-judge --judge-model ${{ inputs.judge_model || 'gemma3:4b' }}"
+          else
+            # llm mode
+            JUDGE_FLAGS="--judge-model ${{ inputs.judge_model || 'gemma3:4b' }}"
+          fi
+
+          echo "Judge mode: ${{ inputs.judge_mode || 'simple' }}"
+          echo "Judge flags: $JUDGE_FLAGS"
+
           # Progress goes to stderr (visible), JSON results go to file
-          npm run --silent dev -- run --suite runtime --no-llm --output json > /tmp/runtime-results.json || true
+          npm run --silent dev -- run --suite runtime $JUDGE_FLAGS --output json > /tmp/runtime-results.json || true
 
           echo "--- JSON Results ---"
           cat /tmp/runtime-results.json