Improve `GroupBy` transformation method names

Now we have a bunch of `GroupBy` transformation methods, but their names are really confusing:

* `sortByGroup`, `sortByCount`, `sortByKey` looks like `sortBy` shortcuts, but actually `sortBy` sorts rows inside groups while others sort the order of groups.
* `filter`/`add` filters rows / add columns to groups, but can be understood as methods that filters/adds new groups (key-group pairs) to the `GroupBy`.
* `updateGroups` - fine, its name perfectly reflects what it does! And this is the only method with the name that clearly makes it clear what it’s working with. In fact, `sortBy`, `filter` and `add` are shortcuts for `updateGroups`, but their names don't reflect that!

So, we need to rename these methods for clear understanding whether method works with key-group pairs or group rows.

Start by reviewing the GroupBy transformation methods named in the issue: sortBy, sortByGroup, sortByCount, sortByKey, filter, add, and updateGroups. Establish a naming scheme that distinguishes operations on group rows from operations on key-group pairs, then trace the affected API usage. Done means the renamed methods consistently communicate which level they transform.

Written by the indexing model from the issue text.