Compare Host and Origin case-insensitively (#3437)

[chrono-glitch]

Fixes #3437

Problem

TransportSecurityMiddleware._validate_host() and _validate_origin() compare
the incoming Host / Origin header against allowed_hosts / allowed_origins
case-sensitively — both the exact-match check and the :* wildcard-port loop.

Host names are case-insensitive (RFC 9110 §4.2.3), and every WHATWG-URL based
client (fetch, undici, browsers, and therefore mcp-remote) lowercases the URL
host before sending. On Windows this is the default path into the bug:
%COMPUTERNAME% is uppercase, so deriving allowed_hosts from the machine name
yields MYHOST:*, the client sends host: myhost:8000, and the server returns
421 to every request while configured exactly as intended.

Fix

Normalize both sides to lowercase in _validate_host and _validate_origin
(exact match and the wildcard-port loop). The original header value is kept in
the logger.warning line for observability. Origin scheme + host are likewise
case-insensitive (RFC 6454 §4); an origin has no path component so lowercasing
the whole string is safe.

Tests

tests/server/test_transport_security.py:

• 4 new parametrized cases in test_validate_request_checks_host_then_origin
  (uppercase request vs lowercase allowlist, host + origin, exact + wildcard).
• new test_validate_host_case_insensitive_with_uppercase_allowlist — the
  Windows scenario: allowed_hosts=["MYHOST:*"] accepts myhost:8000 and
  MYHOST:8000, still rejects other:8000.

Local: uv run pytest tests/server/test_transport_security.py tests/server/test_sse_security.py tests/server/test_streamable_http_security.py
→ 71 passed. ruff check, ruff format --check, pyright clean.

Notes

• Targets main (v2). Happy to open the equivalent against the v1 branch.
• Assisted by an AI coding tool; the change and tests were reviewed, run, and
  are owned 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 #3437.

If a maintainer assigns you to #3437, 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.