[js/webgpu] allow uint8 tensors for webgpu (#19545)

### Description
allow uint8 tensors for webgpu

Author: fs-eire

js/common/lib/tensor-impl.ts

@@ -103,7 +103,7 @@ export class Tensor implements TensorInterface {
         }
         case 'gpu-buffer': {
           if ((type !== 'float32' && type !== 'float16' && type !== 'int32' && type !== 'int64' && type !== 'uint32' &&
-               type !== 'bool')) {
+               type !== 'uint8' && type !== 'bool')) {
             throw new TypeError(`unsupported type "${type}" to create tensor from gpu buffer`);
           }
           this.gpuBufferData = arg0.gpuBuffer;

js/common/lib/tensor.ts

@@ -135,7 +135,7 @@ export declare namespace Tensor {
   /**
    * supported data types for constructing a tensor from a WebGPU buffer
    */
-  export type GpuBufferDataTypes = 'float32'|'float16'|'int32'|'int64'|'uint32'|'bool';
+  export type GpuBufferDataTypes = 'float32'|'float16'|'int32'|'int64'|'uint32'|'uint8'|'bool';
 
   /**
    * represent where the tensor data is stored

js/web/lib/wasm/wasm-common.ts

@@ -169,7 +169,8 @@ export const logLevelStringToEnum = (logLevel?: 'verbose'|'info'|'warning'|'erro
  * Check whether the given tensor type is supported by GPU buffer
  */
 export const isGpuBufferSupportedType = (type: Tensor.Type): type is Tensor.GpuBufferDataTypes => type === 'float32' ||
-    type === 'int32' || type === 'int64' || type === 'bool' || type === 'float16' || type === 'uint32';
+    type === 'float16' || type === 'int32' || type === 'int64' || type === 'uint32' || type === 'uint8' ||
+    type === 'bool';
 
 /**
  * Map string data location to integer value