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/musicgen/encodec_model.rs

@@ -1,6 +1,7 @@
 use crate::nn::{conv1d, conv1d_weight_norm, Conv1d, Conv1dConfig, VarBuilder};
 use anyhow::Result;
 use candle::{DType, IndexOp, Tensor};
+use candle_nn::Module;
 
 // Encodec Model
 // https://github.com/huggingface/transformers/blob/main/src/transformers/models/encodec/modeling_encodec.py

candle-examples/examples/musicgen/musicgen_model.rs

@@ -4,6 +4,7 @@ use crate::nn::{
 use crate::{encodec_model, t5_model};
 use anyhow::Result;
 use candle::{DType, Device, Tensor, D};
+use candle_nn::Module;
 
 // https://github.com/huggingface/transformers/blob/cd4584e3c809bb9e1392ccd3fe38b40daba5519a/src/transformers/models/musicgen/configuration_musicgen.py#L83
 #[derive(Debug, Clone, PartialEq)]

candle-examples/examples/musicgen/t5_model.rs

@@ -4,6 +4,7 @@
 use crate::nn::{embedding, linear, Dropout, Embedding, HiddenAct, Linear, VarBuilder};
 use anyhow::Result;
 use candle::{DType, Tensor, D};
+use candle_nn::Module;
 use std::sync::Arc;
 
 #[derive(Debug, Clone, PartialEq)]