fix(api): keep the base path of the server URL in Twirp requests

[ANDROIDHASSAN]

Since the region-failover change (#733), TwirpClient.request builds every request URL from self._origin, which is scheme://host[:port] only, so a base path in the server URL is dropped. Before #733 the URL was f"{self.host}/{self.prefix}/..." and kept it. A deployment reached through a reverse proxy under a path now gets its API calls sent to the wrong place:

LiveKitAPI(url="https://proxy.example.com/livekit", ...)
# requests go to https://proxy.example.com/twirp/livekit.RoomService/ListRooms
# expected       https://proxy.example.com/livekit/twirp/livekit.RoomService/ListRooms

The first attempt now uses self.host (which keeps the path); fallback regions from /settings/regions are still bare origins, as before. self.host itself also had a doubled slash (https://proxy.example.com//livekit), because the path from urlparse already starts with /; it is now https://proxy.example.com/livekit. For a URL without a path, self.host equals the origin, so the request URL is unchanged.

tests/api/test_twirp_client.py records the URL request() posts to. The two cases with a base path fail on main and pass with this change; the two without a path pass on both. pytest tests/api: 14 passed, 28 skipped (the failover tests need the mock server, which CI starts). ruff format --check, ruff check and mypy livekit-api/livekit pass.

Scope: failover only engages for *.livekit.cloud hosts; region discovery (/settings/regions) and fallback attempts still use bare origins, which is unchanged.

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.

[lukasIO]

thanks!