The auto-generated docs for OTel types from api_data.py may not be enough to explain the overall picture of how the different subsets of statistics cursors overlap. For example, it needs to be clear what happens if we:

• Open with subset fast, OTel type "counter"
• Open with "full", and OTel type "none"
• Open with "full" and OTel type "counter"

And whatever other permutations. Avoid a case bash, and cover it from first principles.