Implement `ed_df.groupby([...]).indices`

- Implement `.indices` similar to pandas
Example:
```python
>>> pd_ecommerce.groupby(["currency","type"]).indices
{('EUR', 'order'): array([ 0, 1, 2, ..., 4672, 4673, 4674], dtype=int64)}
>>> pd_flights.groupby(["Cancelled"]).indices
{False: array([ 0, 1, 2, ..., 13056, 13057, 13058], dtype=int64), True: array([ 3, 8, 12, ..., 13037, 13048, 13051], dtype=int64)}
```
How this works in Pandas: We return a dictionary where values are the positions of that document in the entire index.

Start by tracing the DataFrame groupby implementation behind `ed_df.groupby([...]).indices` and compare its behavior with the pandas examples in the issue. Confirm how grouped rows map to positions in the full index, then verify that `.indices` returns the expected dictionary for single- and multi-column grouping. No source file or test is named, and the issue has not had recent activity.

Written by the indexing model from the issue text.