Quantized GGUF style (#1523)

* Metal quantized modifications proposal.

- Add a device param, wherever needed.
- Create new QMetal storage thing that implements QuantizedType.
- Update everywhere needed.

Fix Python.

Fixing examples.

Fix: fmt + clippy + stub.

Moving everything around.

Only missing the actual implems.

Fixing everything + adding dequantized kernels.

More work.

Fixing matmul.

Fmt + Clippy

Some clippy fixes.

Working state.

Q2K Metal -> Bugged (also present in GGML).
Q4K CPU -> Bugged (present previously, new test catch it).
Q5K CPU -> Bugged (present previously).
Q8_1 Both -> Never really implemented it seems
Q8K metal -> Never implemented in metal

Fixing Q2K bug (present in ggml).

* Cleanup.

* Fix the rebase.

* Removing the fences speeds everything up and *is* correct this time...

* Cleanup the fence.

* After rebase.

* Bad code removal.

* Rebase after phi2 merge + fix replit default to CPU.

* Making the CI happy.

* More happy tests.

---------

Co-authored-by: Nicolas Patry <nicolas@Nicolass-MacBook-Pro.local>

candle-examples/examples/mistral/main.rs

@@ -244,13 +244,14 @@ fn main() -> Result<()> {
 
     let start = std::time::Instant::now();
     let config = Config::config_7b_v0_1(args.use_flash_attn);
+    let device = candle_examples::device(args.cpu)?;
     let (model, device) = if args.quantized {
         let filename = &filenames[0];
-        let vb = candle_transformers::quantized_var_builder::VarBuilder::from_gguf(filename)?;
+        let vb =
+            candle_transformers::quantized_var_builder::VarBuilder::from_gguf(filename, &device)?;
         let model = QMistral::new(&config, vb)?;
-        (Model::Quantized(model), Device::Cpu)
+        (Model::Quantized(model), device)
     } else {
-        let device = candle_examples::device(args.cpu)?;
         let dtype = if device.is_cuda() {
             DType::BF16
         } else {