[Impeller] moved the kernel size assert to a more appropriate location. (#50880)

This is the more appropriate location for the assert now.  This is important if we want to fiddle with the downsample amount now the the kernel is effectively half as big.

[C++, Objective-C, Java style guides]: https://github.com/flutter/engine/blob/main/CONTRIBUTING.md#style

impeller/entity/contents/filters/gaussian_blur_filter_contents.cc

diff --git a/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc b/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc
index 9df1ed5..93d5dd7 100644
--- a/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc
+++ b/impeller/entity/contents/filters/gaussian_blur_filter_contents.cc
@@ -173,9 +173,11 @@
                 linear_sampler_descriptor));
         GaussianBlurVertexShader::BindFrameInfo(
             pass, host_buffer.EmplaceUniform(frame_info));
+        KernelPipeline::FragmentShader::KernelSamples kernel_samples =
+            LerpHackKernelSamples(GenerateBlurInfo(blur_info));
+        FML_CHECK(kernel_samples.sample_count < kMaxKernelSize);
         GaussianBlurFragmentShader::BindKernelSamples(
-            pass, host_buffer.EmplaceUniform(
-                      LerpHackKernelSamples(GenerateBlurInfo(blur_info))));
+            pass, host_buffer.EmplaceUniform(kernel_samples));
         return pass.Draw().ok();
       };
   if (destination_target.has_value()) {
@@ -478,7 +480,6 @@
   KernelPipeline::FragmentShader::KernelSamples result;
   result.sample_count =
       ((2 * parameters.blur_radius) / parameters.step_size) + 1;
-  FML_CHECK(result.sample_count < kMaxKernelSize);
 
   // Chop off the last samples if the radius >= 3 where they account for < 1.56%
   // of the result.