fix: Update response_time for cached results in HybridBackend

The HybridBackend was returning the original response_time from when
the result was first cached, causing all cached queries to show the
same response time (e.g., 519.24ms) regardless of actual cache lookup speed.

This fix measures the actual cache lookup time and updates response_time
when returning cached results, showing the true performance (~1-2ms).

Before:
- Cached query: 519.24ms (always the same, from original query)

After:
- Cached query: 0.7-2ms (varies, actual cache lookup time)

This affects all backends using HybridBackend pattern:
- retraction_watch
- doaj
- crossref_analyzer
- openalex_analyzer

[AI-assisted]

Author: florath

src/aletheia_probe/backends/base.py

@@ -218,16 +218,20 @@ def __init__(self, cache_ttl_hours: int = 24):
 
     async def query(self, query_input: QueryInput) -> BackendResult:
         """Check cache first, then query live API if needed."""
+        start_time = time.time()
+
         # Generate cache key for this query
         cache_key = self._generate_cache_key(query_input)
 
         # Try cache first
         cached_result = get_cache_manager().get_cached_assessment(cache_key)
         if cached_result:
             # Update the cached result to indicate it came from cache
+            cache_lookup_time = time.time() - start_time
             for backend_result in cached_result.backend_results:
                 if backend_result.backend_name == self.get_name():
                     backend_result.cached = True
+                    backend_result.response_time = cache_lookup_time
                     backend_result.data = {**backend_result.data, "from_cache": True}
                     return backend_result