fix(llm): OpenCode Go session header and Go/Zen path detection

[kgrontis]

Summary

OpenCode Go rejects chat requests without a stable x-opencode-session header (HTTP 400), and Odysseus never sent it — plus _detect_provider() could never return opencode-go/opencode-zen because _host_match() is hostname-only while zen/zen/go are path segments. This adds host+path Go/Zen detection, injects an Odysseus/1.0 User-Agent with the per-conversation session id (fallback odysseus-global for probes/model listing), routes Go responses-models (gpt-5.6-luna, grok-4.6, muse-spark) to .../v1/responses with an OpenAI Responses payload, and adds regression tests.

Target branch

• This PR targets dev, not main. All PRs land in dev; main is curated by the maintainer at each release. If your PR is on main by accident, click "Edit" on this PR and change the base.

Linked Issue

Fixes #6281. Related to #4450.

Type of Change

• Bug fix (non-breaking — fixes a confirmed issue)
• New feature (non-breaking — adds new behaviour)
• Breaking change (changes or removes existing behaviour)
• Refactor / cleanup (behaviour unchanged)
• Documentation only
• CI / tooling / configuration

Checklist

• I searched open issues and open PRs — this is not a duplicate.
• This PR targets dev
• My changes are limited to the scope described above — no unrelated refactors or whitespace changes mixed in.
• I actually ran the app (docker compose up or uvicorn app:app) and verified the change works end-to-end. Type-checks and unit tests are not enough.
• I did not run the app/runtime validation and stated that gap in How to Test. Leave this unchecked when the app-run box above is checked.

How to Test

1. Add a Model Endpoint with base URL https://opencode.ai/zen/go/v1 and a valid Go API key.
2. Send any chat message (e.g. hey) on a Go model such as longcat-2.0 — before this fix it fails with 400 ... missing x-opencode-session, after it the request carries User-Agent: Odysseus/1.0 and x-opencode-session: <chat session id>.
3. Automated: run pytest tests/test_opencode_go_responses.py tests/test_opencode_go_headers.py tests/test_provider_detection_detect.py tests/test_provider_detection_host_match.py tests/test_provider_detection_builders.py tests/test_provider_classification.py tests/test_copilot.py tests/test_llm_core_ollama.py -q (from the repo venv) — 129 passed. Also python -m py_compile src/llm_core.py src/endpoint_resolver.py routes/model_routes.py.
4. Gap: I did not run the full app (docker compose up / uvicorn) or a live Go key end-to-end — no Go subscription key available in this environment — so live verification against opencode.ai/zen/go is still needed by a reviewer with a key.
5. Already tried live: gpt-5.6-luna and Muse Spark (muse-spark) previously returned Go 500 Internal server error because Odysseus sent all Go traffic to .../v1/chat/completions while Go serves those models on .../v1/responses (https://opencode.ai/docs/go/#endpoints). This PR now routes them to /responses with a Responses payload (reusing the validated Responses SSE path); live re-verification with a Go key is still needed. The /v1/messages Go models (qwen, minimax) remain out of scope.

Visual / UI changes — REQUIRED if you touched anything that renders

No UI/rendering changes (backend src/ + tests/ only).

• Screenshot or short clip of the change in the running app, attached below. Mobile screenshot too if the change affects mobile.
• Style match: the change uses Odysseus's existing visual language.
• No new component patterns.
• I am not an LLM agent submitting a bulk PR. If you are, please open an issue describing the problem first — bulk auto-generated PRs that don't match the project's visual style are closed on sight, even when the underlying fix is correct.

Screenshots / clips

N/A — no visual change.

[github-actions]

⚠️ PR description is complete; validation evidence is still outstanding

Changed-file classification: backend/runtime.

Author-reported runtime / visual state

• The author explicitly reports that app/runtime validation was not performed.

Checkboxes are author attestations. GitHub Actions results remain the execution evidence for CI; this check does not prove that a local command ran.

---

This comment updates automatically when the description or changed files change.