Refactor request handling with library-agnostic RequestData

[adamtheturtle]

Introduce RequestData dataclass as a frozen, library-agnostic representation of HTTP requests, replacing the requests library's PreparedRequest in the handler layer.

Changes:

• Add RequestData dataclass with method, path, headers, body fields
• Update all handler signatures to accept RequestData instead of PreparedRequest
• Move body normalization (None/str/bytes handling) to adapter boundary in decorators.py
• Remove duplicate _body_bytes() helpers from both API modules
• Decouple core business logic from requests library dependency

This enables future adapter support (respx, httpx, etc) and eliminates type-checking conflicts between mypy, pyrefly, and ruff.

Fixes #2974

---

Note

Medium Risk
Changes the core interface between the responses adapter and all mock route handlers, so incorrect request normalization (body/path/headers) could break request validation or JSON parsing across endpoints.

Overview
Refactors the mock request handling layer to use a new frozen RequestData dataclass (method, path, headers, body) instead of requests.PreparedRequest.

MockVWS._wrap_callback now converts a PreparedRequest into RequestData (including normalizing body to bytes) before invoking route handlers, and both Web Services and Web Query API handlers are updated to consume RequestData, removing duplicated body-parsing helpers and switching from path_url to path.

^{Written by Cursor Bugbot for commit def934d. This will update automatically on new commits. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}

[cursor]

Query API body normalization behavior silently changed

Low Severity

The PR describes the two removed _body_bytes() helpers as "duplicate," but they had different behavior for str bodies. The old query API version returned b"" for string bodies (treating them same as None), while the services API version encoded them via .encode("utf-8"). The unified adapter in _wrap_callback now always encodes str to UTF-8, silently changing the query API's body handling semantics. If a str body ever reaches the query path, it will now be processed as real content rather than treated as empty.

 

[adamtheturtle]

This behavioral change is intentional. The old query `_body_bytes()` silently discarding string bodies (returning `b""`) was arguably a bug — if a string body arrives, encoding it to UTF-8 is strictly more correct than throwing it away.

In practice this is a no-op: the `responses` library always provides `bytes` or `None` for the query endpoint's multipart form body, so the `str` branch is never reached for query requests.