refactor: extract legacy codec interfaces into zarr-codec

[d-v-b]

🤖 AI text below 🤖

Start the extraction of the existing codec API into an independently packaged zarr-codec distribution. zarr_codec.legacy reproduces the current codec base classes, partial-IO mixins, sync capabilities, and CodecPipeline, preserving signatures and inherited behavior. This provides a namespace for the existing contracts while future API designs can coexist separately.

Context for this extraction:

• zarr-python#4217 — fully separate sync and async internal APIs explicitly proposes zarr-storage and zarr-codec subpackages that first reproduce the existing APIs, followed by new interfaces and a bridged migration. It also connects this work to modular array backends and zarrista.
• zarr-python#4317 — how to integrate subpackages proposes testing subpackages against Zarr first, adopting compatible runtime replacements next, and introducing new features before planning major API changes.
• zarr-python#3867 — should the codec API be defined upstream explains why external codec implementations create a circular dependency when Zarr also wants to depend on them. The extraction proposal in the discussion spells out preserving today's API in an initial package release, then developing and testing its successor separately.
• zarr-python#3913 — upstream basic metadata classes and #4268 — subpackage for chunk key encodings give the broader packaging rationale: narrowly scoped components that other Python projects can reuse without depending on the whole array implementation.

These are design and migration references; this draft is an initial extraction and does not complete the proposed API redesign or remove the remaining dependency on zarr.

This is an initial extraction draft: shared foundations still come from zarr, so the package temporarily depends on it. Zarr must not depend on this distribution until that dependency is removed. The copied classes have distinct identities, and Zarr's existing entry points do not yet recognize them as their own nominal base classes. Concrete implementations and Zarr runtime imports are unchanged; there are no deprecations or new interfaces.

Includes package documentation, wheel/sdist configuration, signature-parity and behavioral tests, and package-specific CI for Python 3.12/3.14, including tests against built wheels.

Validation: 15 package tests pass independently on Python 3.12 against the current fork main; the original combined extraction passed all 35 tests against built wheels on Python 3.12 and 3.14. Ruff, workflow lint, and commit hooks pass.