fix: clippy ScalarValue::as_f32→to_f32_scalar + CI infra fixes

- ScalarValue trait: as_f32() → to_f32_scalar() (clippy wrong_self_convention)
- CI template: test-gate uses ci-builder (CUDA) instead of ci-builder-cpu
- VPC: enabled DefaultRoutePropagation for gateway DHCP
- PVCs: all set to Retain reclaim policy
- Argo CLI: configured server mode for archived log retrieval

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

crates/ml-core/src/cuda_autograd/gpu_tensor.rs

@@ -34,19 +34,19 @@ fn raw_ptr_mut(slice: &mut CudaSlice<f32>, stream: &CudaStream) -> u64 {
 
 /// Helper trait for scalar values that can be converted to f32.
 pub trait ScalarValue {
-    fn as_f32(self) -> f32;
+    fn to_f32_scalar(self) -> f32;
 }
 
 impl ScalarValue for f32 {
-    fn as_f32(self) -> f32 { self }
+    fn to_f32_scalar(self) -> f32 { self }
 }
 
 impl ScalarValue for f64 {
-    fn as_f32(self) -> f32 { self as f32 }
+    fn to_f32_scalar(self) -> f32 { self as f32 }
 }
 
 impl ScalarValue for &[f32] {
-    fn as_f32(self) -> f32 { self.first().copied().unwrap_or(0.0) }
+    fn to_f32_scalar(self) -> f32 { self.first().copied().unwrap_or(0.0) }
 }
 
 /// A GPU-resident f32 tensor with shape information.