Add pydantic classes for codecs

[dstansby]

I found myself wanting to specify codecs in v3 array specs, and not wanting to write the JSON by hand, so I ended up down this rabbit hole and implemented pydantic classes for all the codecs in the spec. I'm not sure if this is welcome or duplicates work elsewhere, so I will not be offended if this is not merged :)

[d-v-b]

this looks good to me! I don't think this restricts us any any way, so 👍

[dstansby]

@d-v-b I'm returning to this, and don't want to conflict with what's alread in pydantic-zarr for v3 configuration. In the current v3 code you used a TypedDict for NamdedConfig, and not a pydantic BaseModel, which I what I tried to use here with the codecs. Do you remember why you used TypedDicts instead of pydantic models for the config?

[d-v-b]

@d-v-b I'm returning to this, and don't want to conflict with what's alread in pydantic-zarr for v3 configuration. In the current v3 code you used a TypedDict for NamdedConfig, and not a pydantic BaseModel, which I what I tried to use here with the codecs. Do you remember why you used TypedDicts instead of pydantic models for the config?

I don't remember exactly! It's possible I was thinking about using typeddicts exported from zarr-python, and also I think typeddicts are more ergonomic for very simple metadata than pydantic models (because you don't need to import a new class to use them).

[d-v-b]

We can drop the typeddict annotation if that makes things easier for you! The basemodel representation is definitely useful for validation.

[dstansby]

Okay! Let me have a bit of a think and remember what I was trying to achieve here (easier construction of codec configuration), and I'll loop back with what I think is the right thing to do 👍