Set "max dimensions" to None

`usm_ndarray` can support potentially arbitrary dimensions

Author: ndgrigorian

dpctl/tensor/_array_api.py

@@ -75,7 +75,7 @@ def __init__(self):
         self._capabilities = {
             "boolean indexing": True,
             "data-dependent shapes": True,
-            "max dimensions": 64,
+            "max dimensions": None,
         }
         self._all_dtypes = {
             "bool": dpt.bool,
@@ -108,7 +108,7 @@ def capabilities(self):
                 Value: ``True``
             ``max dimensions``:
                 integer indication the maximum array dimension supported by ``dpctl``.
-                Value: ``64``
+                Value: ``None``
 
         Returns:
             dict:

dpctl/tests/test_tensor_array_api_inspection.py

@@ -74,7 +74,7 @@ def test_array_api_inspection_capabilities():
     capabilities = dpt.__array_namespace_info__().capabilities()
     assert capabilities["boolean indexing"]
     assert capabilities["data-dependent shapes"]
-    assert capabilities["max dimensions"] == 64
+    assert capabilities["max dimensions"] is None
 
 
 def test_array_api_inspection_default_dtypes():