Fix a couple bugs (#1161)

* fix jit reduce for RMS norm

* make strides a single buffer

* better eval error message

* fix compiling with inf and bf16

* fix cpu compile with bf16

mlx/backend/common/make_compiled_preamble.sh

@@ -28,6 +28,7 @@ const char* get_kernel_preamble() {
 return R"preamble(
 $INCLUDES
 $CONTENT
+using namespace mlx::core;
 using namespace mlx::core::detail;
 )preamble";
 }