Add mkl support for matrix multiply. (#86)

* Fix some rebase issues.

* Use mkl instead.

* Use mkl in bert.

* Add the optional mkl feature.

* Conditional compilation based on the mkl feature.

* Add more mkl support.

candle-core/examples/cuda_basics.rs

@@ -1,3 +1,6 @@
+#[cfg(feature = "mkl")]
+extern crate intel_mkl_src;
+
 use anyhow::Result;
 use candle::{Device, Tensor};