Abort OCSP requests when the response times out

## Problem

An OCSP responder can send headers or part of the body, then stop responding. After 30 seconds, `response.buffer()` rejects, but the underlying request is not aborted and its socket can stay open.

`SignedDataVerifier.checkOCSPStatus` uses node-fetch's `timeout: 30000`. That option aborts a request waiting for headers, but does not abort the request when reading the body times out. The body read also sits outside the network-error handler, so the public verification methods return `VERIFICATION_FAILURE` instead of `RETRYABLE_VERIFICATION_FAILURE`.

This affects version 3.1.0 and `main` at `bb0c0f874494321ea2d005329c3dc2188e893d41`.

## Expected behavior

Apply one 30-second deadline to the request and the complete response body. When it expires, abort the request, close the connection and return `RETRYABLE_VERIFICATION_FAILURE`.

Non-timeout response-body and OCSP validation errors should keep their current classification.

## Related work

#345 introduced the OCSP timeout and retryable network-error status. #424 and #425 concern the App Store Server API client; they do not cover the OCSP request in `SignedDataVerifier`.

Start at SignedDataVerifier.checkOCSPStatus and trace the node-fetch request through response.buffer() and the network-error handler. Done means one 30-second deadline aborts the request and complete body read, closes the connection, returns RETRYABLE_VERIFICATION_FAILURE on timeout, and preserves existing classification for other body and OCSP errors.

Written by the indexing model from the issue text.