[Spec change] Clarify accumulator type for reductions and scans

[sycl-issue-bot]

Please review whether KhronosGroup/SYCL-Docs#850 by @Pennycook requires any changes to the CTS.

If changes are required: Open a new PR addressing the changes and link it to this issue.
If no changes are required: Close this issue and proceed with the spec PR.

[Pennycook]

Because SYCL limits the data types and binary operators supported by these interfaces, it's non-trivial to write a test for this. I'm honestly not sure whether this is currently observable, because BinaryOp must return a specific type. There might be a combination of built-in types and specific values that exposes this, but I'm having trouble figuring it out.