Make stats endpoint respect selected time window instead of hardcoded last hour

P99 latency and active count now use the same from/to parameters as the
timeseries sparklines, so all stat cards are consistent with the user's
selected time range.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

cameleer3-server-core/src/main/java/com/cameleer3/server/core/search/ExecutionStats.java

@@ -3,7 +3,7 @@ package com.cameleer3.server.core.search;
 /**
  * Aggregate execution statistics.
  *
- * @param p99LatencyMs 99th percentile duration in milliseconds (last hour)
+ * @param p99LatencyMs 99th percentile duration in milliseconds within the requested time window
  * @param activeCount  number of currently running executions
  */
 public record ExecutionStats(long p99LatencyMs, long activeCount) {}

cameleer3-server-core/src/main/java/com/cameleer3/server/core/search/SearchEngine.java

@@ -28,9 +28,11 @@ public interface SearchEngine {
     /**
      * Compute aggregate stats: P99 latency and count of currently running executions.
      *
+     * @param from start of the time window
+     * @param to   end of the time window
      * @return execution stats
      */
-    ExecutionStats stats();
+    ExecutionStats stats(java.time.Instant from, java.time.Instant to);
 
     /**
      * Compute bucketed time-series stats over a time window.

cameleer3-server-core/src/main/java/com/cameleer3/server/core/search/SearchService.java

@@ -32,8 +32,8 @@ public class SearchService {
     /**
      * Compute aggregate execution stats (P99 latency, active count).
      */
-    public ExecutionStats stats() {
-        return engine.stats();
+    public ExecutionStats stats(java.time.Instant from, java.time.Instant to) {
+        return engine.stats(from, to);
     }
 
     /**