This use case is different as you woudl like to group data first by some attributes and then count them. Simple counting by commas without any further context of each item in the field won’t work.
First, you should import a field that distinguishes blocking subtasks from the rest of sub-taks. Then you can move forward to group them. If blocking information is hidden in a free format in other fields, like issue summary, then you might want to create a new JavaScript calcaulted custom field that check summary for the name pattern and divides all issues in “Blocking” and “Non-blocking” subtasks. And import this new field as a dimension so you could group data by it.
Check out this community post on how to create new dimension from Summary:
Check out this community posts on counting subtasks and grouping them by some attributes: