fix: eagerly validate conversation on service-api and explore chat endpoints

[he-yufeng]

Summary

Follow-up to #37224, which fixed the "hang on non-existing conversation_id" bug (#37197) on the web chat-messages endpoint. The merged commit (31a50a3b20) only changed controllers/web/completion.py, although its description also listed the Service API and console endpoints. As a result the same hang still reproduces on two other chat entry points:

• controllers/service_api/app/completion.py → ChatApi
• controllers/console/explore/completion.py → ChatApi

Both pass a user-supplied conversation_id straight into the streaming generator, where ConversationNotExistsError is raised lazily — after the controller's try/except has already returned the SSE response — so an invalid id hangs on ping events instead of returning 404.

This adds the same eager ConversationService.get_conversation() pre-check before AppGenerateService.generate() on both endpoints. The lookup is identical to the one the generate pipeline already runs at entry (core/app/apps/chat/app_generator.py — same app_model / user / session=db.session() arguments), so valid conversations are unaffected; it only moves the existing check earlier so the error surfaces synchronously and is mapped to 404.

controllers/console/app/completion.py (the debugger ChatMessageApi) was also named in #37224 but routes through _create_chat_message, which already validates the id for AGENT apps and uses a different error-mapping path; it's left as a separate follow-up (tracked in #38800) to keep this change focused and low-risk.

Fixes #38800

Screenshots

N/A — backend error-handling change (an invalid conversation_id now returns 404 instead of hanging on SSE ping events).

Test plan

• Added test_invalid_conversation_id_fails_fast_as_not_found to both endpoints' unit tests: an invalid conversation_id returns 404, and AppGenerateService.generate is never reached (assert_not_called).
• Existing behavior is untouched: the check is guarded by if payload.conversation_id:, so the no-id and valid-id paths are unchanged.
• Ran locally on the touched files: ruff check clean; both controller test suites pass (74 passed). I did not run the full make type-check locally; CI will cover it.

Checklist

• This change requires a documentation update
• I understand that this PR may be closed in case there was no previous discussion or issues. (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.
• I've updated the documentation accordingly.
• I ran make lint && make type-check (backend) and cd web && pnpm exec vp staged (frontend) to appease the lint gods

From Claude Code

[github-actions]

Pyrefly Diff

base → PR

--- /tmp/pyrefly_base.txt	2026-07-12 12:46:04.422424881 +0000
+++ /tmp/pyrefly_pr.txt	2026-07-12 12:45:54.078256717 +0000
@@ -2253,7 +2253,7 @@
 ERROR `None` is not subscriptable [unsupported-operation]
    --> tests/unit_tests/controllers/service_api/app/test_audio.py:186:16
 ERROR Argument `list[dict[str, Any]] | None` is not assignable to parameter `obj` with type `Sized` in function `len` [bad-argument-type]
-   --> tests/unit_tests/controllers/service_api/app/test_completion.py:154:20
+   --> tests/unit_tests/controllers/service_api/app/test_completion.py:155:20
 ERROR Argument value `Literal[0]` violates Pydantic `ge` constraint `Literal[1]` for field `limit` [bad-argument-type]
   --> tests/unit_tests/controllers/service_api/app/test_conversation.py:82:35
 ERROR Argument value `Literal[101]` violates Pydantic `le` constraint `Literal[100]` for field `limit` [bad-argument-type]

[github-actions]

Pyrefly Type Coverage

Metric	Base	PR	Delta
Type coverage	52.85%	52.85%	+0.00%
Strict coverage	52.36%	52.36%	+0.00%
Typed symbols	32,803	32,808	+5
Untyped symbols	29,539	29,541	+2
Modules	2988	2988	0