Update on "[ET-VK] Implemement linear_dq8ta_q4gsw"

Title says it all!

Build upon the support for quantized linear introduced in the previous diffs to enable dynamically quantized linear.

Also included in this diff is a cleanup of the glslh files used across quantized linear implementations.

Differential Revision: [D81931060](https://our.internmc.facebook.com/intern/diff/D81931060/)

[ghstack-poisoned]

Author: ssjia

backends/vulkan/patterns/quantized_linear.py

@@ -181,6 +181,9 @@ def is_input_dynamic_perchannel_quantized(self) -> bool:
         if self.quantize_input_node is None:
             return False
 
+        if not isinstance(self.input_scales_node, torch.fx.Node):
+            return False
+
         # For dynamic quantization, input scale node should be a getitem operator
         # retrieving the output of a choose_qparams op
         if self.input_scales_node.target != operator.getitem: