refactor(duckdb): deprecate read_in_memory

This functionality is available via `create_table` (and also available
across all backends instead of just DuckDB).

ibis/backends/duckdb/__init__.py

@@ -851,8 +851,12 @@ def _read_parquet_pyarrow_dataset(
         # by the time we execute against this so we register it
         # explicitly.
 
+    @util.deprecated(
+        instead="Pass in-memory data to `create_table` instead.",
+        as_of="9.1",
+        removed_in="10.0",
+    )
     def read_in_memory(
-        # TODO: deprecate this in favor of `create_table`
         self,
         source: pd.DataFrame
         | pa.Table

ibis/backends/tests/test_register.py

@@ -277,7 +277,9 @@ def test_register_iterator_parquet(
     assert table.count().execute()
 
 
-# TODO: modify test to use read_in_memory when implemented xref: 8858
+# TODO: remove entirely when `register` is removed
+# This same functionality is implemented across all backends
+# via `create_table` and tested in `test_client.py`
 @pytest.mark.notimpl(["datafusion"])
 @pytest.mark.notyet(
     [
@@ -311,7 +313,9 @@ def test_register_pandas(con):
     assert t.x.sum().execute() == 6
 
 
-# TODO: modify test to use read_in_memory when implemented xref: 8858
+# TODO: remove entirely when `register` is removed
+# This same functionality is implemented across all backends
+# via `create_table` and tested in `test_client.py`
 @pytest.mark.notimpl(["datafusion", "polars"])
 @pytest.mark.notyet(
     [