speed up `FindBestThresholdFromHistogram` (#2867)

* speed up for const hessian

* rename template

* some refactorings

* refine

* refine

* simplify codes

* fix random in feature histogram

* code refine

* refine

* try fix

* make gcc happy

* remove timer

* rollback some changes

* more templates

* fix a bug

* reduce the cost of timer

* fix gpu

* fix bug

* fix gpu

windows/LightGBM.vcxproj

@@ -237,7 +237,6 @@
     <ClInclude Include="..\include\LightGBM\dataset.h" />
     <ClInclude Include="..\include\LightGBM\dataset_loader.h" />
     <ClInclude Include="..\include\LightGBM\feature_group.h" />
-    <ClInclude Include="..\include\LightGBM\json11.hpp" />
     <ClInclude Include="..\include\LightGBM\lightgbm_R.h" />
     <ClInclude Include="..\include\LightGBM\meta.h" />
     <ClInclude Include="..\include\LightGBM\metric.h" />
@@ -250,6 +249,7 @@
     <ClInclude Include="..\include\LightGBM\utils\array_args.h" />
     <ClInclude Include="..\include\LightGBM\utils\common.h" />
     <ClInclude Include="..\include\LightGBM\utils\file_io.h" />
+    <ClInclude Include="..\include\LightGBM\utils\json11.h" />
     <ClInclude Include="..\include\LightGBM\utils\log.h" />
     <ClInclude Include="..\include\LightGBM\utils\openmp_wrapper.h" />
     <ClInclude Include="..\include\LightGBM\utils\pipeline_reader.h" />

windows/LightGBM.vcxproj.filters

@@ -201,9 +201,6 @@
     <ClInclude Include="..\include\LightGBM\utils\file_io.h">
       <Filter>include\LightGBM\utils</Filter>
     </ClInclude>
-    <ClInclude Include="..\include\LightGBM\json11.hpp">
-      <Filter>include\LightGBM\utils</Filter>
-    </ClInclude>
     <ClInclude Include="..\src\treelearner\cost_effective_gradient_boosting.hpp">
       <Filter>src\treelearner</Filter>
     </ClInclude>
@@ -213,6 +210,9 @@
     <ClInclude Include="..\src\io\multi_val_sparse_bin.hpp">
       <Filter>src\io</Filter>
     </ClInclude>
+    <ClInclude Include="..\include\LightGBM\utils\json11.h">
+      <Filter>include\LightGBM\utils</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="..\src\application\application.cpp">