Arm backend: Add tutorial for dynamic range 8x8. (pytorch#22930)

Adds a VGF tutorial for dynamic W8A8 quantization.

The tutorial covers:

* configuring `VgfQuantizer` for dynamic INT8 activations and static
INT8 weights;
* preparing and converting a model with PT2E;
* lowering the quantized model to VGF;
* generating and running the resulting `.pte`;
* applying the flow to DeiT-Tiny.

It also explains the dynamic activation qparam flow and the INT8 MATMUL
→ INT32 accumulation → FP32 rescaling path.

Assisted with Codex.

cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils
@Sebastian-Larsson @robell @rascani

Signed-off-by: Elena Zhelezina <elena.zhelezina@arm.com>

Author: wwwind

docs/source/backends/arm-vgf/arm-vgf-overview.md

@@ -140,6 +140,8 @@ described in the rest of this guide but with a concrete end-to-end sample.
 
 **→{doc}`/backends/arm-vgf/tutorials/vgf-getting-started` — Getting started tutorial.**
 
+**→{doc}`/backends/arm-vgf/tutorials/vgf-dynamic-w8a8-quantization` — Dynamic W8A8 quantization tutorial.**
+
 **→{doc}`/backends/arm-vgf/VGF_op_support` — VGF supported operators.**
 
 
@@ -152,5 +154,6 @@ arm-vgf-partitioner
 arm-vgf-quantization
 arm-vgf-troubleshooting
 tutorials/vgf-getting-started
+tutorials/vgf-dynamic-w8a8-quantization
 VGF_op_support
 ```

docs/source/backends/arm-vgf/arm-vgf-quantization.md

@@ -13,6 +13,8 @@ The quantization schemes supported by the VGF Backend are:
 
 Weight-only quantization is not currently supported on the VGF backend.
 
+For an end-to-end example of configuring and exporting dynamic W8A8 Linear layers, see the [Dynamic W8A8 Quantization Tutorial](tutorials/vgf-dynamic-w8a8-quantization.md).
+
 ### Partial Quantization
 
 The VGF backend supports partial quantization, where only parts of the model