Thrift SDK explain() silently drops GROUP BY and HAVING

**Description**

`table.output(...).group_by(...).having(...).explain()` over the thrift SDK explains a different query than the one that was built: `group by` and `having` never reach the server.

In `python/infinity_sdk/infinity/remote_thrift/table.py`, `_explain_query` hardcodes `group_by_list=None` when calling `conn.explain`, and `remote_thrift/client.py`'s `explain()` has no `having_expr` parameter at all - even though `ExplainRequest` in the thrift IDL has both `group_by_list` and `having_expr` fields and the server reads them. The plain `select` path forwards both correctly; only `explain` was left behind.

**Reproduction**

```python
import infinity
from infinity.table import ExplainType

inf = infinity.connect("http://localhost:23817")
db = inf.get_database("default_db")
t = db.create_table("t", {"c1": {"type": "varchar"}, "c2": {"type": "float"}})
t.insert({"c1": "a", "c2": 1.0})

res = t.output(["c1", "sum(c2)"]).group_by(["c1"]).having("sum(c2) > 0").explain(ExplainType.Physical)
print(res)
```

The physical plan contains no `Aggregate` operator - the explained statement lost its `group by` / `having`. Calling `.to_result()` on the same builder forwards both correctly, so this is specific to `explain()`. It can also be confirmed without a server by mocking the connection: `_explain_query` passes `group_by_list=None` no matter what the query holds.

Start in python/infinity_sdk/infinity/remote_thrift/table.py at _explain_query, then compare its connection call with the forwarding used by the plain select path. Read remote_thrift/client.py explain() and the ExplainRequest fields in the thrift IDL. Use a mocked connection to verify that group_by and having reach explain, and confirm the reproduced physical plan retains the aggregate query.

Written by the indexing model from the issue text.