Update df.ww.value_counts to work with all column types

Currently `df.ww.value_counts` only returns counts for columns that have a `category` semantic tag added. This method might be useful for other columns as well. We could remove this restriction and if users only wanted to get the value counts for categorical columns they could preselect the columns of interest using the `select` method prior to calling `value_counts`:

```python
df.ww.select(include='category').value_counts()
```