Static-frame

Frame.iter_tuple_items()`` now exposes a ``constructor`` argument to control creation of axis containers.

Added ``Batch.apply_items``.

Added ``Frame.count``, ``Series.count``, ``Batch.count``.

Added ``Frame.sample``, ``Series.sample``, ``Index.sample``, ``IndexHierarchy.sample``, ``Batch.sample``.

Added ``Frame.via_T`` and ``IndexHierarchy.via_T`` accessors for opposite axis binary operator application of 1D operands.

IndexHierarchy.iter_label`` now defaults to iterating full depth labels.

``Batch.__repr__()`` is no longer a display that exhausts the stored generator.

``Frame.iter_tuple()`` now exposes a ``constructor`` argument to control creation of axis containers.

Fixed issue in using ``Frame.extend`` with zero-length ``Frame``.

Implemented ``Frame.isin`` on ``TypeBlocks``.

Implemented ``Frame.clip`` on ``TypeBlocks``.

Series.from_element`` now works correctly with tuples.

``Batch`` element handling now avoids diagonal formations; ``Batch.apply()`` now handles elements correctly.

``dtypes`` parameters can now be provided with ``dict_values`` instances.

``Frame.to_parquet``, ``Frame.to_arrow`` now convert ``np.datetime64`` units to nanosecond if not supported by PyArrow.

Bus`` now exposes ``max_persist`` parameter to define the maximum number of loaded ``Frame`` retained by the ``Bus``.

Added ``len()`` to ``via_str`` interfaces.

``Frame.iter_element`` now takes an ``axis`` argument to determine element order, where 0 is row major, 1 is column major.

Silenced ``NaturalNameWarning`` via ``tables`` in ``StoreHDF5``.

``StoreSQLite`` will now re-write, rather than update, a file path where an SQLite DB already exists.

Improved handling for iterating zero-sized ``Frame``.

Improved type detection when performing operations on ``Frame.iter_element`` iterators.

``Frame.shift()`` ``file_value`` parameter is now key-word argument only.

``Frame.roll()`` ``include_index``, ``include_columns`` is now key-word argument only.