perf(ebpf): improve decodeEvents (HashTaskID)

Metrics from local benchmarking on a AMD Ryzen 9 7950X 16-Core Processor
with 64GB RAM running 6.12.25-1-MANJARO. The stress test was run
with 8mi events triggered in 8 threads.

- CPU time

| Function     | Before    | After     | Change    | Improvement (%) |
|--------------|-----------|-----------|-----------|-----------------|
| decodeEvents | 26.87 sec | 25.22 sec | -1.65 sec | 6.14%           |
| HashTaskID   | 0.38 sec  | 0.34 sec  | -0.04 sec | 10.53%          |

- Memory

| Metric        | Function     | Before  | After   | Change   | Improv. (%) |
|---------------|--------------|---------|---------|----------|-------------|
| alloc_space   | decodeEvents | 8.52 GB | 8.48 GB | -0.04 GB | 0.47%       |
| inuse_objects | decodeEvents | 1.17 K  | 0.85 K  | -0.32 K  | 27.35%      |
| alloc_objects | decodeEvents | 91.96 M | 89.31 M | -2.65 M  | 2.88%       |

Author: geyslan

pkg/ebpf/events_pipeline.go

@@ -275,7 +275,11 @@ func (t *Tracee) decodeEvents(ctx context.Context, sourceChan chan []byte) (<-ch
 			evt.Metadata = nil
 			// compute hashes using normalized times
 			evt.ThreadEntityId = utils.HashTaskID(eCtx.HostTid, uint64(evt.ThreadStartTime))
-			evt.ProcessEntityId = utils.HashTaskID(eCtx.HostPid, traceetime.BootToEpochNS(eCtx.LeaderStartTime))
+			if eCtx.HostTid == eCtx.HostPid && eCtx.StartTime == eCtx.LeaderStartTime {
+				evt.ProcessEntityId = evt.ThreadEntityId // leader thread - same as thread entity id
+			} else {
+				evt.ProcessEntityId = utils.HashTaskID(eCtx.HostPid, traceetime.BootToEpochNS(eCtx.LeaderStartTime))
+			}
 			evt.ParentEntityId = utils.HashTaskID(eCtx.HostPpid, traceetime.BootToEpochNS(eCtx.ParentStartTime))
 
 			// If there aren't any policies that need filtering in userland, tracee **may** skip