Preserve primary key when loading models

[sakshichitnis27]

Fixes #828

Summary

Preserve the requested primary key when loading HashModel and JsonModel records with a missing stored primary key.

Added tests for default and custom primary keys.

Testing

• pytest -q tests/test_primary_key_loading.py tests_sync/test_primary_key_loading.py

---

Note

Low Risk
Narrow change to get() hydration only; uses the same pk already passed to the API and is covered by targeted tests. Stored PK values in the document, if present, would be overwritten by the requested key—intended for legacy missing-PK data.

Overview
HashModel.get and JsonModel.get now set the primary key from the lookup argument when Redis returns a document that omits the PK field (e.g. legacy hashes/JSON without pk stored).

Before validation, the loaders merge {cls._meta.primary_key.name: pk} into the fetched payload on both the normal and byte-decode paths for hashes, and on the JSON path for JsonModel. Custom primary keys (e.g. external_id) use the configured field name, so .pk and the custom field stay aligned with the key used in get(pk).

New async tests in tests/test_primary_key_loading.py cover missing-PK hash/JSON documents, undecoded hash bytes, and a custom Field(primary_key=True).

^{Reviewed by Cursor Bugbot for commit bc45387. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 2 potential issues.

^{Reviewed by Cursor Bugbot for commit eea94cf. Configure here.}