refactor: Use Pydantic models instead of custom dict representations (#650)

## Summary

- Replace hand-written `_representations.py` helper functions (302 lines
deleted) with Pydantic request models (`UpdateActorRequest`,
`ScheduleCreate`, `WebhookCreate`, `WebhookUpdate`, etc.) for
constructing API payloads.
- Simplify `_update()` / `_create()` in `ResourceClient` to accept
`**kwargs` and use a new `_clean_json_payload()` static method on
`ResourceClientBase` instead of the removed `filter_none_values()`
utility.
- Remove unused `enum_to_value()` and `filter_none_values()` from
`_utils.py` — Pydantic serialization handles enum conversion and `None`
filtering.
- Add `populate_by_name=True` to all generated Pydantic model configs
via `datamodel-codegen` configuration, so models can be initialized
using either `snake_case` field names or `camelCase` aliases.
- Models had to be adjusted, see PR to apify-docs -
apify/apify-docs#2291.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Author: vdusek

pyproject.toml

@@ -206,6 +206,7 @@ wrap_string_literal = true
 snake_case_field = true
 use_subclass_enum = true
 extra_fields = "allow"
+allow_population_by_field_name = true
 aliases = "datamodel_codegen_aliases.json"
 formatters = ["ruff-check", "ruff-format"]
 

src/apify_client/_consts.py

@@ -25,7 +25,6 @@
 DEFAULT_MIN_DELAY_BETWEEN_RETRIES = timedelta(milliseconds=500)
 """Default minimum delay between retries."""
 
-
 DEFAULT_WAIT_FOR_FINISH = timedelta(seconds=999999)
 """Default maximum wait time for job completion (effectively infinite)."""