Use bindgen-cuda for the custom-kernel example. (#1536)

* Use bindgen-cuda for the custom-kernel example.

* Only depend on the kernels when cuda is enabled.

* Skip rustfmt.

candle-examples/examples/custom-ops/main.rs

@@ -6,7 +6,8 @@
 #[cfg(feature = "mkl")]
 extern crate intel_mkl_src;
 
-#[allow(unused)]
+#[rustfmt::skip]
+#[cfg(feature = "cuda")]
 mod cuda_kernels;
 
 use clap::Parser;