Match `ww.init()` behaviour between DataFrames and Series

Currently calling `df.ww.init()` infers the logical types and performs transformations as well. Doing the same for `series.ww.init()` only infers the logical type, but doesn't change the underlying `dtype` or transform the data as per [here](https://woodwork.alteryx.com/en/latest/start.html?highlight=init_series#Using-Woodwork-with-a-Series). We use `ww.init_series()` to perform that functionality.

I think a user would reasonably expect `ww.init()` to result in the same operation regardless of whether they're using it on a dataframe or series. This issue covers flipping the behaviour between `ww.init_series()` and `series.ww.init()`.