Add print function with nonzero dim args (#844)

* Add print overload with non-zero dimension args

Add a new matx::print() implementation that accepts one or more
arguments to specify the number of elements to print in the given
dimension.

* Document behavior of a dimension being set to 0 in print()

* Change unit test to generic TypeType type

Author: tbensonatl

include/matx/core/print.h

@@ -736,6 +736,21 @@ namespace matx {
     cuda::std::apply([&](auto &&...args) { fprint(stdout, op, args...); }, tp);
   }
 
+  /**
+   * @brief Print all of a tensor's values to stdout
+   *
+   * This form of `print()` is a specialization for 1D+ tensors. A size of zero in
+   * dimension prints all elements in that dimension.
+   *
+   * @tparam Op Operator input type
+   * @param op Operator input
+   */
+  template <typename Op, typename... Args,
+            std::enable_if_t<(Op::Rank() > 0 && sizeof...(Args) > 0), bool> = true>
+  void print(const Op &op, [[maybe_unused]] Args... dims) {
+    fprint(stdout, op, dims...);
+  }
+
   /**
    * @brief Print a tensor's all values to stdout
    *

test/00_operators/OperatorTests.cu

@@ -4745,9 +4745,12 @@ TYPED_TEST(OperatorTestsFloatAllExecs, Print)
   MATX_ENTER_HANDLER();
   using TestType = cuda::std::tuple_element_t<0, TypeParam>;
 
-  auto t = make_tensor<TestType>({3});
-  auto r = ones<TestType>(t.Shape());
+  auto t1 = make_tensor<TestType>({3});
+  auto r1 = ones<TestType>(t1.Shape());
+  print(r1);
+
+  auto t3 = matx::make_tensor<TestType>({3, 2, 20});
+  print(matx::ones(t3.Shape()), 1, 0, 2);
 
-  print(r);
   MATX_EXIT_HANDLER();
 }