fix(search): bound and police the scholarly metadata lookups

The arXiv and OpenAlex title resolvers called two hardcoded endpoints with a
hand-written "Odysseus/0.20" User-Agent, no outbound-URL policy, and a full
12s timeout each. APP_VERSION was already 1.0.3, so the agent string was wrong
the moment it was written, and a scholarly query could hold a user-facing
search open for the sum of all three hops.

Move the endpoints to named constants, build the User-Agent from APP_VERSION,
run both calls through check_outbound_url (they follow redirects, so the final
host is not the one in the constant), and give the whole SearXNG -> OpenAlex ->
arXiv chain one shared wall-clock budget.

The budget is a ContextVar rather than a parameter so the existing test doubles
for _openalex_title_results and _arxiv_title_results keep working unchanged.

src/constants.py

@@ -140,6 +140,18 @@ LLM_HOSTS = [h.strip() for h in os.getenv("LLM_HOSTS", "").split(",") if h.strip
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
 SEARXNG_INSTANCE = os.getenv("SEARXNG_INSTANCE", "http://localhost:8080")
 
+# Scholarly title resolution. These are the only third-party metadata APIs the
+# search path calls directly, so they get named constants rather than literals
+# repeated at each call site. The budget bounds the whole SearXNG -> OpenAlex ->
+# arXiv chain: each hop used to get its own full timeout, so one scholarly query
+# could stall a user-facing search for the sum of all three.
+ARXIV_API_URL = "https://export.arxiv.org/api/query"
+OPENALEX_API_URL = "https://api.openalex.org/works"
+SCHOLARLY_LOOKUP_TIMEOUT = float(os.getenv("ODYSSEUS_SCHOLARLY_LOOKUP_TIMEOUT", "12"))
+SCHOLARLY_LOOKUP_TOTAL_BUDGET = float(
+    os.getenv("ODYSSEUS_SCHOLARLY_LOOKUP_BUDGET", "20")
+)
+
 # Cleanup configuration
 CLEANUP_ENABLED = os.getenv("CLEANUP_ENABLED", "True").lower() == "true"
 CLEANUP_INTERVAL_HOURS = int(os.getenv("CLEANUP_INTERVAL_HOURS", "24"))