Fix status-cache poisoning, add HTTP timeouts, coerce custom status names

- _status_choices no longer caches failures: a transient /ticket_form_fields
  error used to cache [] forever, silently degrading unresolved=True to an
  unfiltered listing (including Resolved/Closed) until container restart.
  unresolved=True and list_ticket_statuses now return an explicit error when
  the status list can't be fetched instead of degrading silently.
- All API calls (server and skill helper) now use a 30s timeout with a clean
  error, so a hung connection can't block a tool call indefinitely.
- _coerce_status falls back to the instance's live status list, so custom
  names like "Working" coerce to their id instead of 400ing at the API.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

scripts/fs.py

@@ -93,7 +93,7 @@ def request(method: str, path: str, params: dict[str, str], data: str | None) ->
     req = urllib.request.Request(url, data=body, method=method.upper(), headers=headers)
 
     try:
-        with urllib.request.urlopen(req) as resp:
+        with urllib.request.urlopen(req, timeout=30) as resp:
             raw = resp.read().decode()
             out = {"ok": True, "status": resp.status,
                    "data": json.loads(raw) if raw else {}}
@@ -116,6 +116,8 @@ def request(method: str, path: str, params: dict[str, str], data: str | None) ->
                 "method": method.upper(), "url": url}
     except urllib.error.URLError as e:
         return {"ok": False, "error": f"Network error: {e.reason}", "url": url}
+    except TimeoutError:
+        return {"ok": False, "error": "Request timed out after 30s", "url": url}
 
 
 def main() -> None: