feature(threading): use explicit/configurable thread pool executor

[jhamman]

Previously, Zarr was using a home rolled to_thread function to offload work to an implicit executor. This PR changes this in three main ways:

1. Use asyncio.to_thread instead of the home rolled version
2. Setup a configurable ThreadPoolExecutor alongside the IOthread and IOloop. The number of threads can be configured using threading.max_workers.
3. Provide a teardown function (zarr.core.sync.cleanup_resources) to explicitly cleanup resources created by Zarr at runtime. This function is executed at shutdown (via atexit) and can also be triggered manually (e.g. in the dask test suite)

TODO:

• Add unit tests and/or doctests in docstrings
• Add docstrings and API docs for any new/modified user-facing classes and functions
• New/modified features documented in docs/tutorial.rst
• Changes documented in docs/release.rst
• GitHub Actions have all passed
• Test coverage is 100% (Codecov passes)

[d-v-b]

looks like a nice improvement!