Improve documentation for stream argument to __dlpack__ (#185)

Closes gh-183 (SYCL related, don't require an integer for `stream`)

The need to document the ownership of `stream` came up in pytorch/pytorch#57781.

Author: rgommers

spec/API_specification/array_object.md

@@ -432,9 +432,15 @@ Exports the array for consumption by {ref}`function-from_dlpack` as a DLPack cap
 
     -   array instance.
 
--   **stream**: _Optional\[ int ]_
+-   **stream**: _Optional\[ Union\[ int, Any ]]_
 
-    -   a Python integer representing a pointer to a stream. `stream` is provided by the consumer to the producer to instruct the producer to ensure that operations can safely be performed on the array. The pointer must be a positive integer or `-1`. If `stream` is `-1`, the value may be used by the consumer to signal "producer must not perform any synchronization". Device-specific notes:
+    -   for CUDA and ROCm, a Python integer representing a pointer to a stream, on devices that support streams. `stream` is provided by the consumer to the producer to instruct the producer to ensure that operations can safely be performed on the array (e.g., by inserting a dependency between streams via "wait for event"). The pointer must be a positive integer or `-1`. If `stream` is `-1`, the value may be used by the consumer to signal "producer must not perform any synchronization". The ownership of the stream stays with the consumer.
+
+        On CPU and other device types without streams, only `None` is accepted.
+
+        For other device types which do have a stream, queue or similar synchronization mechanism, the most appropriate type to use for `stream` is not yet determined. E.g., for SYCL one may want to use an object containing an in-order `cl::sycl::queue`. This is allowed when libraries agree on such a convention, and may be standardized in a future version of this API standard.
+
+        Device-specific notes:
 
         :::{admonition} CUDA
         -   `None`: producer must assume the legacy default stream (default).