Merge branch 'develop' into conv_splitk_f32

device_operation/include/convolution_forward_specialization.hpp

+#ifndef CONVOLUTION_FORWARD_SPECIALIZATION
+#define CONVOLUTION_FORWARD_SPECIALIZATION
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+
+enum ConvolutionForwardSpecialization_t
+{
+    Default,
+    Filter1x1Pad0,
+    Filter1x1Stride1Pad0,
+    OddC,
+};
+
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck
+#endif