Hdmf

Minor improvements
- Updated ``MultiContainerInterface.__make_add`` to clarify the docstring for ``add_*`` methods generated by the function. oruebel ([758](https://github.com/hdmf-dev/hdmf/pull/758))
- Support "allow_none=True" in docval for args with non-None default. rly ([757](https://github.com/hdmf-dev/hdmf/pull/757))

Bug fixes
- Fixed deploy release CI. rly ([759](https://github.com/hdmf-dev/hdmf/pull/759))

Bug fixes
- Fixed deploy release CI. rly ([754](https://github.com/hdmf-dev/hdmf/pull/754))

Minor improvements
- Allow manual triggering of some GitHub Actions. rly ([744](https://github.com/hdmf-dev/hdmf/pull/744))
- Relaxed input validation of `HDF5IO` to allow for s3fs support. Existing arguments of `HDF5IO` are modified as follows: i) `mode` was given a default value of "r", ii) `path` was given a default value of `None`, and iii) `file` can now accept an `S3File` type argument. bendichter ([746](https://github.com/hdmf-dev/hdmf/pull/746))
- Added ability to create and get back handle to empty HDF5 dataset. ajtritt ([747](https://github.com/hdmf-dev/hdmf/pull/747))
- Added `AbstractContainer._in_construct_mode` that is set and modified only by the ObjectMapper when constructing an
object from a builder read from a file. Subclasses of `AbstractContainer` can check `_in_construct_mode`
during the initialization phase as part of ``__init__`` to distinguish between actions during construction
(i.e., read from disk) vs. creation by the user, e.g., to determine whether to raise a warning or error when
encountering invalid data to support reading and correcting data that is invalid while preventing creation
of new data that is invalid. rly ([751](https://github.com/hdmf-dev/hdmf/pull/751))

Bug fixes
- Fixed PyNWB dev CI. rly ([749](https://github.com/hdmf-dev/hdmf/pull/749))

Bug fixes
- Fix error message when using ``GenericDataChunkIterator`` and chunk shape is larger than dataset size. rly (743)
- Fix automatic class generation of extension data types that inherit from both another data type and
``MultiContainerInteface``. rly (741)

Bug fixes
- Fixed release deployment CI. rly (734, 735, 736)

New features
- Added utility functions and classes useful for implementation of I/O backend to ease maintainability and reuse of functionality across I/O backends. oruebel (697)
- Added `HDF5IODataChunkIteratorQueue` class in `hdmf.backends.hdf5.h5_utils` as a new helper class for managing the write of `DataChunkIterator`s to HDF5. oruebel (697)
- Added new `hdmf/io/utils.py` module with new I/O utility classes: 1) `WriteStatusTracker` as a simple data structure for tracking the write status of `Builders`, and 2) `NamespaceToBuilderHelper` to help with converting of a namespace to a `Builder` for I/O. oruebel (697)
- Added `get_min_bounds` function to `hdmf.data_utils.DataChunk`. This functionality was originally part of `HDF5IO.__write_chunk__()` and has been moved here to enable reuse of the code across data backends. oruebel (697)
- Added `ignore_string_to_byte` option for `TestCase.assertContainerEqual` to allow us to ignore conversion from string to bytes and just compare unicode values, as different backends may store strings differently. oruebel (697)
- Allow `zarr.Array` as array datatype in docval to support conversion data stored in Zarr to HDMF HDF5 data. oruebel (721)
- Allow `hdmf.common.table.DynamicTable.add_column` to accept nested lists as column data and automatically create the `VectorIndex` and `VectorData` objects required to represent the ragged array. oruebel (728)
- Added a warning when the ``__init__`` of a class mapped to an HDMF-common data type or an autogenerated data type class
is passed positional arguments instead of all keyword arguments. rly (730)
- Added helper function `hdmf.utils.popargs_to_dict` for popping docval args into a dict. rly (732)

Bug fixes
- Fixed error with modifying files that contain external links to other files (e.g., shallow copies). rly (709)
- Fixed opening of files in append mode on Windows when the files contain links to other open files. rly (710)
- Updated `HDF5IO` to always set the `location` attribute of `GroupBuilders`, `DatasetBuilders`, and `LinkBuilders` on read. oruebel (697)
- Updated `HDF5IO.get_types` to correctly determine the data type for `bytes` data. oruebel (697)

Minor improvements
- Updated `HDF5IO` to use the new `WriteStatusTracker`, `NamespaceToBuilderHelper`, and `HDF5IODataChunkIteratorQueue` helper classes. oruebel (697)
- Updated `hdmf.common.sparse.CSRMatrix` to avoid direct dependency on h5py as a particular storage backend. oruebel (697)
- Improved readability of ``Container`` code. rly (707)
- Use GitHub Actions for all CI. rly (718)
- Allow `write_args=None` to be passed to `HDF5IO.export`. rly (733)
- Updated requirements and fixed minor documentation issues and warnings. rly (731)

Test enhancements
- Moved test functions to ease reuse and updated tests accordingly. oruebel (697)
- Moved `Baz`, `BazData`, `BazCpdData`, `BazBucket`, `get_baz_buildmanager` test data classes from `tests.unit.test_io_hdf5_h5tools` to `tests.unit.utils` to ease reuse and updated tests accordingly. Also `_get_baz_manager` was renamed to `get_baz_buildmanager` as part of this move. oruebel (697)
- Added numerous tests to `tests/unit/common/test_sparse.py` to enhance testing of the `CSRMatrix` type. oruebel (697)

Documentation and tutorial enhancements:
- Add copy button to code blocks. weiglszonja oruebel (726)