fix build err inbuild with minimal_build conjuncting disable_exceptio…

…ns flags (#14524)

### Description
If we set flag 'disable_exceptions' to build ORT:


`onnxruntime/contrib_ops/cpu/quantization/qlinear_global_average_pool.cc.o`
woundn't generate such symbols which used by qlinear_pool.c
```
0000000000000000 W _ZN11onnxruntime7contrib27ComputeQLinearGlobalAvgPoolIaEENS_6common6StatusEPKT_fS4_PS4_fS4_lllbPNS_11concurrency10ThreadPoolE
0000000000000000 W _ZN11onnxruntime7contrib27ComputeQLinearGlobalAvgPoolIhEENS_6common6StatusEPKT_fS4_PS4_fS4_lllbPNS_11concurrency10ThreadPoolE
```
so we get a error of undefined symbols of
ComputeQLinearGlobalAvgPool<uin8_t> and
ComputeQLinearGlobalAvgPool<in8_t>......


### Motivation and Context
<!-- - Why is this change required? What problem does it solve?
- If it fixes an open issue, please link to the issue here. -->

onnxruntime/contrib_ops/cpu/quantization/qlinear_global_average_pool.cc

@@ -55,6 +55,38 @@ Status ComputeQLinearGlobalAvgPool(
   return Status::OK();
 }
 
+// GCC's unexplained behavior:
+// GCC wouldn't generate corresponding symbols versus function instances below when "--disable-exceptions"
+// and "--minimal-build" are combined on linux build.
+// But this two symbols are required by qlinear_pool.cc.
+// The other compilers wouldn't hit it and works fine, and we also didn't see it in the other platforms, such as Android.
+// So we are doing explicit instantiation here for every compilers/platforms happy.
+template Status ComputeQLinearGlobalAvgPool<int8_t>(
+    const int8_t* x,
+    float x_scale,
+    int8_t x_zero_point,
+    int8_t* y,
+    float y_scale,
+    int8_t y_zero_point,
+    int64_t N,
+    int64_t C,
+    int64_t image_size,
+    bool channels_last,
+    concurrency::ThreadPool* tp);
+
+template Status ComputeQLinearGlobalAvgPool<uint8_t>(
+    const uint8_t* x,
+    float x_scale,
+    uint8_t x_zero_point,
+    uint8_t* y,
+    float y_scale,
+    uint8_t y_zero_point,
+    int64_t N,
+    int64_t C,
+    int64_t image_size,
+    bool channels_last,
+    concurrency::ThreadPool* tp);
+
 Status QLinearGlobalAveragePool::Compute(OpKernelContext* context) const {
   const auto tensor_x_scale = context->Input<Tensor>(1);
   const auto tensor_x_zero_point = context->Input<Tensor>(2);