Summary

Request._prepare() already keeps an explicit Content-Length when the content encoder provides a default Transfer-Encoding. This applies the same precedence in the other direction: an explicit Transfer-Encoding now takes precedence over a body-derived Content-Length.

This makes handling of explicit framing headers consistent for fixed and streaming request content, and adds regression coverage for fixed content.

WSGITransport buffers request bodies before constructing the WSGI environment. When a buffered body has no explicit Content-Length, the transport now exposes its known length as CONTENT_LENGTH so WSGI applications can consume the input normally.

Validation

• pytest tests/httpx2/models/test_requests.py tests/httpx2/test_wsgi.py -q — 38 passed
• pytest tests/httpx2 -q — 1721 passed, 1 skipped
• scripts/check — passed

Checklist

• I understand that this PR may be closed in case there was no previous discussion. (This doesn't apply to typos!)
• I've added a test for each change that was introduced, and I tried as much as possible to make a single atomic change.
• No documentation update is needed for this internal request-preparation consistency change.