Add support for ascending list in `df.sort_values`

[stress-tess]

Add support for something like this:

df.sort_values(by=['col_a', 'col_b'], ascending=[True, False])

Example of a workaround provided for user

>>> pd_df = pd.DataFrame({'col_a': (np.arange(5) % 2 == 0).astype(int), 'col_b': np.arange(5,0,-1), 'col_c': np.arange(5)})

>>> ak_df = ak.DataFrame(pd_df)

>>> def list_sort_values_workaround(df, col_names, order_list):
     return df[ak.coargsort([df[c_name] if ascend else -df[c_name] for c_name, ascend in zip(col_names, order_list)])]

[stress-tess]

I think we basically just add an if isinstance(ascending, list) then do a size check. If those are both fine then return the line from the workaround

[stress-tess]

it's also worth noting this only applies to numeric values. Right now we don't guarantee sorting on strings, categoricals anyway, only ordering