[python] Make dummy classes constructible with any arguments (#4749)

* Make dummy classes constructible with any arguments

* Update compat.py

python-package/lightgbm/compat.py

@@ -14,12 +14,14 @@
     class pd_Series:  # type: ignore
         """Dummy class for pandas.Series."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass
 
     class pd_DataFrame:  # type: ignore
         """Dummy class for pandas.DataFrame."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass
 
     concat = None
     is_dtype_sparse = None
@@ -52,7 +54,8 @@ class pd_DataFrame:  # type: ignore
     class dt_DataTable:  # type: ignore
         """Dummy class for datatable.DataTable."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass
 
 
 """sklearn"""
@@ -143,19 +146,23 @@ class _LGBMRegressorBase:  # type: ignore
     class Client:  # type: ignore
         """Dummy class for dask.distributed.Client."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass
 
     class dask_Array:  # type: ignore
         """Dummy class for dask.array.Array."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass
 
     class dask_DataFrame:  # type: ignore
         """Dummy class for dask.dataframe.DataFrame."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass
 
     class dask_Series:  # type: ignore
         """Dummy class for dask.dataframe.Series."""
 
-        pass
+        def __init__(self, *args, **kwargs):
+            pass