Add default compressors to config

[brokkoli71]

This PR adds:

• default compressor for zarr_format=2
• specification of default compressors dependent on dtype in zarr.config

fixes #2267

Should _get_default_array_bytes_codec for zarr_format=3 also be configurable in zarr.config?

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)

[brokkoli71]

I am a little confused: in which cases should we use VLenBytes and in which cases VLenUTF8? Or do we need both at the same time?

[brokkoli71]

Also, do we need a default compressor or are default filters sufficient?

[jhamman]

Thanks for working on this @brokkoli71. I think @normanrz is also going to review this but I wanted to also bring up an additional point.

In Zarr2 we told people to set zarr.storage.default_compressor = SomeCompressor() This was simple but also an odd way to manage config. What we have now is much better. However, I wonder if we should do something to catch folks trying do set the default_compressor variable. Thoughts?

[dstansby]

Thanks! Couple of things:

• Print statements need removing
• The config naming could probably be improved

This would also be a good oppurtunity to update or add user facing documentation on what default compressors are used for what types of data - is that something you could add in this PR?

[normanrz]

I think we should also have defaults for v3:

          "v3_default_codecs": {
                "numeric": ["bytes", "zstd"],
                "string": ["vlen-utf8"],
                "bytes": ["vlen-bytes"],
            },

[LDeakin]

I think we should also have defaults for v3:

          "v3_default_codecs": {
                "numeric": ["bytes", "zstd"],
                "string": ["vlen-utf8"],
                "bytes": ["vlen-bytes"],
            },

zstd isn't in the spec yet: zarr-developers/zarr-specs#256

[brokkoli71]

thanks for your feedback, i will integrate it this week 👍🏼

[normanrz]

Thanks! Could you please go through the docstrings again and add some info about the default compressor, filters, codecs? Then, I think this is good to go.

[dstansby]

I've left lots of small requests for changes, mainly around the docstrings - in general they are great and clear, but I think worth fixing a lot of the little issues while we're here.

I left most of the docstring comments in asynchronous.py, but they also apply to the other files that have updated docstrintgs.

[dstansby]

Suggested change

	- For numeric arrays, the default is `BytesCodec` and `ZstdCodec`.
	- For numeric arrays, the default is `BytesCodec` and `ZstdCodec`.

Can we also document the default compression level (and any other parameters) here?

[d-v-b]

is there a reason why none of the default compressors / codecs have a configuration?

[d-v-b]

and a second question, why aren't strings / bytes compressed with zstd by default, like numeric dtypes?