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

@@ -116,7 +116,6 @@ def __init__(self, mm_node: torch.fx.Node) -> None:
 
         # If input is not quantized, then we are done
         if self.quantize_input_node is None:
-            raise Exception("Input is not quantized")
             self.match_found = True
             return
 
@@ -478,7 +477,6 @@ def replace_quantized_linear_patterns(
         and match.is_weight_pergroup_quantized()
         and utils.is_in_4bit_range(weight_tensor)
     ):
-        raise Exception("Unsupported pattern")
         make_linear_q4gsw_op(
             ep, graph_module, match, weight_tensor, weight_scales_tensor
         )

backends/vulkan/test/custom_ops/CMakeLists.txt

@@ -92,7 +92,7 @@ if(TARGET vulkan_backend)
   # Define operator prototypes
   add_operator_prototype(add)
   add_operator_prototype(q8csw_linear)
-  add_operator_prototype(quantized_q4gaw_linear)
-  add_operator_prototype(quantized_int4_linear)
   add_operator_prototype(q8csw_conv2d)
+  add_operator_prototype(q4gsw_linear)
+  add_operator_prototype(choose_qparams_per_row)
 endif()