Add device-aware output dtype for dpt.round() with boolean input (#2851)

This PR proposes device-aware output dtype resolution for
`dpnp.tensor.round()` with `boolean` input to handle devices that do not
support `float16`

Boolean support for round() was originally added in #2817
[6f5a792](6f5a792)
to match NumPy behavior where numpy.round(bool) returns float16 rather
than an integral type like int8.
However on devices without fp16 support, returning float16 is not
viable.

The bool type mapping was removed from the round kernel and an
acceptance
function `_acceptance_fn_round` was added to ensure the fallback in
`_find_buf_dtype`
prefers floating-point output over integral types for boolean input

Result :
fp16 devices: round(bool) -> float16 
non-fp16 devices: round(bool) -> float32

Author: vlad-perevezentsev

dpnp/tensor/_elementwise_funcs.py

@@ -33,6 +33,7 @@
     _acceptance_fn_divide,
     _acceptance_fn_negative,
     _acceptance_fn_reciprocal,
+    _acceptance_fn_round,
     _acceptance_fn_subtract,
     _resolve_weak_types_all_py_ints,
 )
@@ -1723,7 +1724,11 @@
 """
 
 round = UnaryElementwiseFunc(
-    "round", ti._round_result_type, ti._round, _round_docstring
+    "round",
+    ti._round_result_type,
+    ti._round,
+    _round_docstring,
+    acceptance_fn=_acceptance_fn_round,
 )
 del _round_docstring
 

dpnp/tensor/_type_utils.py

@@ -133,6 +133,13 @@ def _acceptance_fn_reciprocal(arg_dtype, buf_dt, res_dt, sycl_dev):
         return True
 
 
+def _acceptance_fn_round(arg_dtype, buf_dt, res_dt, sycl_dev):
+    # for boolean input, prefer floating-point output over integral
+    if arg_dtype.kind == "b" and res_dt.kind != "f":
+        return False
+    return True
+
+
 def _acceptance_fn_subtract(
     arg1_dtype, arg2_dtype, buf1_dt, buf2_dt, res_dt, sycl_dev
 ):
@@ -970,6 +977,7 @@ def _default_accumulation_dtype_fp_types(inp_dt, q):
     "_find_buf_dtype2",
     "_to_device_supported_dtype",
     "_acceptance_fn_default_unary",
+    "_acceptance_fn_round",
     "_acceptance_fn_reciprocal",
     "_acceptance_fn_default_binary",
     "_acceptance_fn_divide",

dpnp/tensor/libtensor/include/kernels/elementwise_functions/round.hpp

@@ -116,7 +116,6 @@ template <typename T>
 struct RoundOutputType
 {
     using value_type = typename std::disjunction<
-        td_ns::TypeMapResultEntry<T, bool, sycl::half>,
         td_ns::TypeMapResultEntry<T, std::uint8_t>,
         td_ns::TypeMapResultEntry<T, std::uint16_t>,
         td_ns::TypeMapResultEntry<T, std::uint32_t>,