Static-frame

Fixed issue in using ``relabel()`` on columns in ``FrameGO``.

Fixed issue in using ``Frame.drop`` with ``IndexHierarchy`` on either axis.

Unified ``to_frame`` and ``to_frame_go`` interfaces on ``Frame``, ``FrameGO``, and ``IndexHierarchy``.

Enabled ``include_index``, ``include_columns`` parameters for ``Frame.to_parquet``.

Added ``columns_select`` parameter to ``Frame.from_parquet``.

Updated requirements: pyarrow==0.16.0

Refined ``Frame.from_arrow`` usage of ChunkedArray, disabling ``date_as_object``, enabling ``self_destruct``, and improving handling of NumPy array extraction.

Added ``STATIC`` attribute to ``ContainerBase`` and all subclasses.

Fixed issue in assigning a column to a ``FrameGO`` from a generator that raises an exception.

Added ``difference`` method to all ``Index`` subclasses.

Added ``index_constructor`` and ``columns_constructor`` parameters to ``Frame.from_pandas``; ``index_constructor`` added to ``Series.from_pandas``.

Refined ``IndexBase.from_pandas``.

Fixed issue introduced into ``Frame.iter_group`` and ``Frame.iter_group_items`` when selecting a single column with an object dtype.

Fixed mapping lookups to use single-argument tuples in ``map_any_iter_items`` and ``map_fill_iter_items`` and related methods.

Improvements to ``any`` and ``all`` methods on all containers when using ``skipna=True`` and NAs are presernt; now, a ``TypeError`` will now be raised when NAs are found and ``skipna=False``.

When converting from Pandas 1.0 extension dtypes, proper NumPy types are used if no ``pd.NA`` are present; if ``pd.NA`` are present, they are replaced with ``np.nan`` in the resulting object array.