feat(burn): wire SIMD sin/cos via hpc::vml (eliminate f64 roundtrip)

float_sin → ndarray::hpc::vml::vssin (F32x16 direct, no f64 conversion)
float_cos → ndarray::hpc::vml::vscos (F32x16 direct, no f64 conversion)

Original burn-ndarray: cast f32→f64, compute sin/cos, cast f64→f32.
Our path: operate directly on f32 via SIMD polynomial approximation.

Total SIMD-wired ops: exp, log, sqrt, abs, sin, cos, sigmoid (7 ops).

30 tests passing.

https://claude.ai/code/session_01Y69Vnw751w75iVSBRws7o7

Author: claude

crates/burn/src/ops/tensor.rs

@@ -561,6 +561,11 @@ where
     }
 
     fn float_cos(tensor: FloatTensor<Self>) -> FloatTensor<Self> {
+        #[cfg(feature = "simd")]
+        let tensor = match try_vml_unary(tensor, ndarray::hpc::vml::vscos) {
+            Ok(result) => return result,
+            Err(t) => t,
+        };
         execute_with_float_dtype!(tensor, FloatElem, |array: SharedArray<FloatElem>| {
             array
                 .mapv_into(|a: FloatElem| (a.to_f64()).cos().elem())
@@ -577,6 +582,11 @@ where
     }
 
     fn float_sin(tensor: FloatTensor<Self>) -> FloatTensor<Self> {
+        #[cfg(feature = "simd")]
+        let tensor = match try_vml_unary(tensor, ndarray::hpc::vml::vssin) {
+            Ok(result) => return result,
+            Err(t) => t,
+        };
         execute_with_float_dtype!(tensor, FloatElem, |array: SharedArray<FloatElem>| {
             array
                 .mapv_into(|a: FloatElem| (a.to_f64()).sin().elem())