[Suggestion] Very and extremely low dimensional data set

The low dimensional data set allows up to 100 columns. This is considered low in most fields. However, in some fields, the normal range of columns is far lower, particularly sports prediction. It is rare to use more than 30 columns in sports prediction. A summary by Google AI suggests 5-30 columns is the standard range for sports prediction models, and based on my experience in this field, this is accurate. Therefore, I propose 2 new data sets:

- Extremely low (10 columns limit)
- Very low (30 columns limit)

It would also be ideal to be able to filter the leaderboard based on groups of subsets. For example, seeing the results for low dimensional + tiny data set would be useful.