fix(ci): bump GPU test opt-level to 2, revert one-time cache clean

opt-level=1 caused DQN lib tests to timeout at the 2-hour deadline
(113 tests × debug-mode GPU init = too slow). opt-level=2 matches
local dev behavior.

Revert the cargo clean -p steps now that stale Candle cache is purged.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

infra/k8s/argo/ci-pipeline-template.yaml

@@ -342,11 +342,6 @@ spec:
 
             export PATH="${CARGO_HOME}/bin:${PATH}"
 
-            # Clean stale incremental cache for crates affected by major refactors
-            # (Candle→cudarc migration invalidates all ML crate artifacts)
-            echo "=== Cleaning stale incremental cache ==="
-            cargo clean -p ml -p ml-core -p ml-dqn -p ml-ppo -p ml-supervised 2>/dev/null || true
-
             echo "=== Waiting for Redis sidecar ==="
             for i in 1 2 3 4 5; do
               if printf 'PING\r\n' | nc -w1 127.0.0.1 6379 2>/dev/null | grep -q PONG; then

infra/k8s/argo/gpu-test-pipeline-template.yaml

@@ -134,7 +134,7 @@ spec:
           - name: LD_LIBRARY_PATH
             value: /usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64
           - name: CARGO_PROFILE_TEST_OPT_LEVEL
-            value: "1"
+            value: "2"
         volumeMounts:
           - name: git-ssh-key
             mountPath: /etc/git-ssh
@@ -208,10 +208,6 @@ spec:
             export PATH="${CARGO_HOME}/bin:${PATH}"
             export CUDA_COMPUTE_CAP={{workflow.parameters.cuda-compute-cap}}
 
-            # --- Clean stale incremental cache (Candle→cudarc migration) ---
-            echo "=== Cleaning stale ML crate cache ==="
-            cargo clean -p ml -p ml-core -p ml-dqn -p ml-ppo -p ml-supervised 2>/dev/null || true
-
             # --- PTX cache invalidation ---
             # Purge stale cached PTX if any CUDA kernel source changed since last run.
             bash scripts/ptx-cache-invalidate.sh "${CARGO_TARGET_DIR}/.ptx_cache"