[SPARK-43502][PYTHON][CONNECT] DataFrame.drop` should accept empty column

### What changes were proposed in this pull request?
Make `DataFrame.drop` accept empty column

### Why are the changes needed?
to be consistent with vanilla PySpark

### Does this PR introduce _any_ user-facing change?
yes

```
In [1]: df = spark.createDataFrame([(1, 21), (2, 30)], ("id", "age"))

In [2]: df.drop()
```

before:
```
In [2]: df.drop()
---------------------------------------------------------------------------
PySparkValueError                         Traceback (most recent call last)
Cell In[2], line 1
----> 1 df.drop()

File ~/Dev/spark/python/pyspark/sql/connect/dataframe.py:449, in DataFrame.drop(self, *cols)
    444     raise PySparkTypeError(
    445         error_class="NOT_COLUMN_OR_STR",
    446         message_parameters={"arg_name": "cols", "arg_type": type(cols).__name__},
    447     )
    448 if len(_cols) == 0:
--> 449     raise PySparkValueError(
    450         error_class="CANNOT_BE_EMPTY",
    451         message_parameters={"item": "cols"},
    452     )
    454 return DataFrame.withPlan(
    455     plan.Drop(
    456         child=self._plan,
   (...)
    459     session=self._session,
    460 )

PySparkValueError: [CANNOT_BE_EMPTY] At least one cols must be specified.
```

after
```
In [2]: df.drop()
Out[2]: DataFrame[id: bigint, age: bigint]
```

### How was this patch tested?
enabled UT

Closes #41180 from zhengruifeng/connect_drop_empty_col.

Authored-by: Ruifeng Zheng <[email protected]>
Signed-off-by: Ruifeng Zheng <[email protected]>

Author: zhengruifeng

python/pyspark/sql/connect/dataframe.py

@@ -445,11 +445,6 @@ def drop(self, *cols: "ColumnOrName") -> "DataFrame":
                 error_class="NOT_COLUMN_OR_STR",
                 message_parameters={"arg_name": "cols", "arg_type": type(cols).__name__},
             )
-        if len(_cols) == 0:
-            raise PySparkValueError(
-                error_class="CANNOT_BE_EMPTY",
-                message_parameters={"item": "cols"},
-            )
 
         return DataFrame.withPlan(
             plan.Drop(

python/pyspark/sql/connect/plan.py

@@ -664,7 +664,8 @@ def __init__(
         columns: List[Union[Column, str]],
     ) -> None:
         super().__init__(child)
-        assert len(columns) > 0 and all(isinstance(c, (Column, str)) for c in columns)
+        if len(columns) > 0:
+            assert all(isinstance(c, (Column, str)) for c in columns)
         self._columns = columns
 
     def plan(self, session: "SparkConnectClient") -> proto.Relation:

python/pyspark/sql/tests/connect/test_parity_dataframe.py

@@ -84,11 +84,6 @@ def test_to_pandas_with_duplicated_column_names(self):
     def test_to_pandas_from_mixed_dataframe(self):
         self.check_to_pandas_from_mixed_dataframe()
 
-    # TODO(SPARK-43502): DataFrame.drop should support empty column
-    @unittest.skip("Fails in Spark Connect, should enable.")
-    def test_drop_empty_column(self):
-        super().test_drop_empty_column()
-
 
 if __name__ == "__main__":
     import unittest