Fix GitHub Actions workflows to upload build libraries and remove LD_LIBRARY_PATH

Changes:
- Update tesla-k80-ci.yml to upload build/lib/ollama/ containing CUDA backend
- Remove all LD_LIBRARY_PATH environment variables (no longer needed with RPATH)
- Test workflows now receive libggml-cuda.so enabling GPU offload

This fixes the issue where test workflows couldn't offload to GPU because the
CUDA backend library wasn't included in the artifact.

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

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

.github/workflows/tesla-k80-multi-gpu-tests.yml

@@ -30,8 +30,6 @@ jobs:
           chmod +x ollama
           ls -lh ollama
           ./ollama --version
-        env:
-          LD_LIBRARY_PATH: /usr/local/cuda-11.4/lib64:/usr/local/cuda-11.4/targets/x86_64-linux/lib:/usr/lib64:/usr/local/lib64
 
       - name: Verify multi-GPU setup
         run: |
@@ -51,21 +49,15 @@ jobs:
           go build -o ../../test-runner .
           cd ../..
           ls -lh test-runner
-        env:
-          LD_LIBRARY_PATH: /usr/local/cuda-11.4/lib64:/usr/local/cuda-11.4/targets/x86_64-linux/lib:/usr/lib64:/usr/local/lib64
 
       - name: Validate multi-GPU test configuration
         run: |
           ./test-runner validate --config test/config/models.yaml --profile multi-gpu
-        env:
-          LD_LIBRARY_PATH: /usr/local/cuda-11.4/lib64:/usr/local/cuda-11.4/targets/x86_64-linux/lib:/usr/lib64:/usr/local/lib64
 
       - name: Run multi-GPU tests
         run: |
           ./test-runner run --profile multi-gpu --config test/config/models.yaml --output test-report-multi-gpu --verbose
         timeout-minutes: 60
-        env:
-          LD_LIBRARY_PATH: /usr/local/cuda-11.4/lib64:/usr/local/cuda-11.4/targets/x86_64-linux/lib:/usr/lib64:/usr/local/lib64
 
       - name: Check multi-GPU test results
         run: |