[scheduler] Early exit cancel path in set_timer_common_ and fix benchmark

Add find_first parameter to cancel_item_locked_ and
mark_matching_items_removed_locked_, defaulting to false (cancel all).

When find_first=true, stops after the first match and exits across
containers. set_timer_common_ passes find_first=true where the
cancel-before-add invariant guarantees at most one match.

The public cancel path uses the default find_first=false to cancel ALL
matches, needed for DelayAction parallel mode where skip_cancel=true
can create multiple items with the same key.

Also fixes the Scheduler_SetInterval benchmark to call
process_to_add() every kKeyCount registrations with a final drain
after the inner loop, reflecting production behavior.

tests/benchmarks/core/bench_scheduler.cpp

@@ -99,11 +99,21 @@ BENCHMARK(Scheduler_SetTimeout);
 static void Scheduler_SetInterval(benchmark::State &state) {
   Scheduler scheduler;
   Component dummy_component;
+  // Number of distinct interval keys; controls how many unique timers exist
+  // simultaneously and the drain cadence for process_to_add().
+  static constexpr int kKeyCount = 5;
 
   for (auto _ : state) {
     for (int i = 0; i < kInnerIterations; i++) {
-      scheduler.set_interval(&dummy_component, static_cast<uint32_t>(i % 5), 1000, []() {});
+      scheduler.set_interval(&dummy_component, static_cast<uint32_t>(i % kKeyCount), 1000, []() {});
+      // Drain to_add_ periodically to reflect production behavior where
+      // process_to_add() runs each main loop iteration. Without this,
+      // cancelled items accumulate in to_add_ causing O(n²) scan cost.
+      if ((i + 1) % kKeyCount == 0) {
+        scheduler.process_to_add();
+      }
     }
+    // Final drain in case kInnerIterations is not a multiple of 5
     scheduler.process_to_add();
     benchmark::DoNotOptimize(scheduler);
   }