Support Half/BFloat16 in nonzero (pytorch#7850)

Partial fix for pytorch#7748.

Author: swolchok

kernels/portable/cpu/op_nonzero.cpp

@@ -88,10 +88,9 @@ Tensor& nonzero_out(KernelRuntimeContext& ctx, const Tensor& in, Tensor& out) {
 
   ET_KERNEL_CHECK(ctx, check_nonzero_args(in, out), InvalidArgument, out);
 
-  ET_SWITCH_REAL_TYPES_AND(
-      Bool, in.scalar_type(), ctx, "nonzero.out", CTYPE, [&] {
-        nonzero<CTYPE>(ctx, in, out);
-      });
+  ET_SWITCH_REALHBBF16_TYPES(in.scalar_type(), ctx, "nonzero.out", CTYPE, [&] {
+    nonzero<CTYPE>(ctx, in, out);
+  });
 
   return out;
 }

kernels/test/op_nonzero_test.cpp

@@ -28,10 +28,8 @@ class OpNonzeroTest : public OperatorTest {
   void test_dtype() {
     TensorFactory<DTYPE> tf_input;
     TensorFactory<ScalarType::Long> tf_long;
-    // clang-format off
-    Tensor a = tf_input.make(/*sizes=*/{2, 2}, /*data=*/{2, 0,
-                                                         2, 4});
-    // clang-format on
+    Tensor a = tf_input.make(
+        /*sizes=*/{2, 2}, /*data=*/{CTYPE(2), CTYPE(0), CTYPE(2), CTYPE(4)});
     Tensor out = tf_long.zeros({3, 2});
 
     op_nonzero_out(a, out);
@@ -45,7 +43,7 @@ class OpNonzeroTest : public OperatorTest {
 
 TEST_F(OpNonzeroTest, AllDtypesSupported) {
 #define TEST_ENTRY(ctype, dtype) test_dtype<ctype, ScalarType::dtype>();
-  ET_FORALL_REAL_TYPES(TEST_ENTRY);
+  ET_FORALL_REALHBBF16_TYPES(TEST_ENTRY);
 #undef TEST_ENTRY
 }