Fail pending send_raw_request waiters when the read stream yields an exception

[chaucerj]

Fixes #1401

What

When a transport's read stream yields an exception item, pending send_raw_request waiters were only woken by the on_stream_exception observer (a no-op by default) — they parked until their own timeout elapsed. Only EOF produced CONNECTION_CLOSED.

_dispatch's exception branch now also fans the raw exception out to every pending waiter. send_raw_request re-raises it as-is, so callers see the transport's original exception type (e.g. httpx.ReadTimeout) instead of a generic timeout. _fan_out_closed is unchanged in behavior (delegates to a generalized _fail_pending).

Why

A faulted read stream means no response can ever arrive; parking the waiters until timeout is indistinguishable from a hang. For transient faults where the stream stays open, later requests keep working.

How verified

• New test: a mid-await transport exception wakes the waiter with that exact exception, a late response is dropped, and a subsequent round-trip succeeds (asyncio + trio).
• Full tests/shared + tests/client suites pass (one pre-existing subprocess-spawn failure on this machine, reproducible on clean main — environment issue, unrelated).
• ruff + pyright clean.

---

This contribution was developed with AI assistance (Claude Code); the approach and code were reviewed and verified by me.

[github-actions]

This PR has been closed automatically. This repo only keeps pull requests open when they come from a maintainer, or from a contributor a maintainer has assigned to the linked issue, and you aren't currently assigned to #1401.

If a maintainer assigns you to #1401, this PR reopens on its own and there's nothing more you need to do here. Assignment is a maintainer call based on capacity; comments that only ask to be assigned don't factor in. What does help is engaging on the issue itself by confirming the repro, explaining why it matters for your use case, or describing the approach you'd take.

You're welcome to keep pushing commits here (just avoid force-pushing, since GitHub can't reopen a rewritten branch), but that on its own won't get the PR reviewed or the issue assigned, and realistically most auto-closed PRs stay closed. There's no need to open a new PR either way.

CONTRIBUTING.md has the full reasoning, but in short:

• We're a small team with very little capacity to review community PRs right now.
• Many recent PRs are AI-generated with little human review, and reviewing one carefully still costs a maintainer as much time as it ever did. A well-described issue is usually more useful to us than the code.

Maintainers: reopen, remove missing-issue-link, or add bypass-issue-check to override.