Populate primary key value for the row just inserted into DBAPI's `lastrowid` and SQLAlchemy's `inserted_primary_key` attributes

### Problem
In SQLAlchemy (ORM) land, the standard is that the cursor object provides `inserted_primary_key` and `lastrowid` attributes. However, currently, both of them do not return anything sensible.
```python
insert_stmt = table.insert(dict(name="John Doe", age=37))
cursor = engine.execute(insert_stmt)
print(cursor.inserted_primary_key)
print(cursor.lastrowid)
```
```
()
None
```

### Motivation
[`dataset`](https://dataset.readthedocs.io/) uses SQLAlchemy's `inserted_primary_key` attribute for returning the pk value from its `insert` method, see https://github.com/pudo/dataset/blob/1.5.2/dataset/table.py#L120-L121. It is important to make this work in order to be able to reference records just inserted, for manipulating them later, or for referencing them elsewhere.

### Other statements
> `.inserted_primary_key` returns a tuple of the primary key of the single inserted row using the `connection.execute()` command. A good reason to use [.inserted_primary_key](https://docs.sqlalchemy.org/en/latest/core/connections.html#sqlalchemy.engine.CursorResult.inserted_primary_key) instead of [.lastrowid](https://docs.sqlalchemy.org/en/latest/core/connections.html#sqlalchemy.engine.CursorResult.lastrowid): `.lastrowid` is database specific, it may work for some databases while not for others. But `.inserted_primary_key` will always work regardless of the underlying database.
>
> -- https://stackoverflow.com/a/51258633