Add a simple Module trait and implement it for the various nn layers (#500)

* Start adding the module trait.

* Use the module trait.

* Implement module for qmatmul.

candle-examples/examples/bert/model.rs

@@ -1,5 +1,5 @@
 use candle::{DType, Device, Result, Tensor};
-use candle_nn::{Embedding, VarBuilder};
+use candle_nn::{Embedding, Module, VarBuilder};
 use serde::Deserialize;
 
 pub const DTYPE: DType = DType::F32;