Define and add `countDistinct` aggregation for `Pivot` and `PivotGroupBy`

[Allex-Nik]

Follow up of #533.

The countDistinct is currently not defined on Pivot and PivotGroupBy. We need to add these overloads for the API consistency (for example, the count function has them) and more straightforward counting distinct values on these receivers.

Functionality to add:

• On Pivot:
  df.pivot { column_0 }.countDistinct()
df.pivot { column_0 }.countDistinct { column_1 }

• On PivotGroupBy:
  df.pivot { column_0 }.groupBy { column_1 }.countDistinct()
df.pivot { column_0 }.groupBy { column_1 }.countDistinct { column_2 }

Tests and KDocs should be added, relevant existing documentation (for countDistinct and pivot) should be updated.

[zaleslaw]

I agree that this use case exists in SQL World, but let's have like a low priority thing

SELECT *
FROM ...
PIVOT (
  COUNT(DISTINCT user_id)
  FOR month IN (...)
)