Static-frame

Set ``arraykit`` version to 0.3.0: replaced usage of ``np.nonzero()`` with ``first_true_1d()`` and ``first_true_2d()``.

Specifiers of dtypes given to ``astype()`` and related interfaces will now permit "object" to be used (in addition to ``object``).

Corrected issue when selecting ``IndexHierarchy`` labels that contain tuples.

``IndexHiearchy.astype()`` now properly delegates the ``name`` attribute.

Better usage of ``np.concatenate`` in ``TypeBlocks.transpose()`` and ``HierarchicalLocMap.build_offsets_and_overflow()``.

Specifiers of dtypes given to ``astype()`` and related interfaces will now raise if NumPy will implicitly convert the argument to ``object``.

``Batch.to_frame()`` now exposes ``index_constructor``, ``columns_constructor`` arguments.

Improvements to usage of ``index_constructor`` and ``columns_constructor`` arguments in ``Frame.from_concat`` when ``Series`` are provided as components.

When using a ``Batch`` to write to an archive, non-unique labels will now raise a ``StoreLabelNonUnique`` exception.

IndexHierarchy.__setstate__`` now properly sets indexers to be immutable.

``Bus`` with associated ``Store`` instances are now pickleable after removing ``_weak_cache`` in ``Store.__getstate__()``.

Series.isna()``, ``Series.dropna()``, and related functions now properly handle arrays that contain objects such as ``Frame`` or ``np.ndarray`` that raise for usage of ``__bool__()``.

Set ``arraykit`` version to 0.2.9: ``isna_element`` now identifies ``pd.Timestamp('nat')`` as a NA value, and invalid ``datetime64`` strings given to ``Frame.from_delimited`` and related interfaces now properly raise exceptions.

Corrected issue in ``IndexHierarchy.label_widths_at_depth()`` which caused incorrect cell merging in XLSX output when ``merge_hierarchical_labels`` is True.