Remove unnecessary bottlenecks

1. No need to wait on join() the monitoring thread, it is safely killed
   when duct process exits.
2. pass when ps fails (because inner process has finished)
3. Collect the first sample immediately rather than waiting for
   <sample-interval>

Fixes #36

src/duct.py

@@ -151,7 +151,7 @@ def collect_sample(self):
                         "timestamp": datetime.now().astimezone().isoformat(),
                     }
         except subprocess.CalledProcessError:
-            process_data["error"] = "Failed to query process data"
+            pass
         return process_data
 
     def write_pid_samples(self):
@@ -199,7 +199,7 @@ def monitor_process(report, process, report_interval, sample_interval):
         totals = report.calculate_total_usage(sample)
         report.update_max_resources(sample, totals)
         report.update_max_resources(report._sample, sample)
-        if report.elapsed_time >= (report.number + 1) * report_interval:
+        if report.elapsed_time >= report.number * report_interval:
             report.write_pid_samples()
             report.update_max_resources(report.max_values, report._sample)
             report._sample = defaultdict(dict)  # Reset sample
@@ -406,7 +406,6 @@ def main():
             target=monitor_process, args=monitoring_args
         )
         monitoring_thread.start()
-        monitoring_thread.join()
 
     if args.record_types in ["all", "system-summary"]:
         report.collect_environment()

test_script.py

@@ -21,8 +21,8 @@ def consume_memory(size):
 
 
 def main(duration, cpu_load, memory_size):
-    consume_memory(memory_size)
     print("this is of test of STDERR: ERRRRRRRRRRRRRRR", file=sys.stderr)
+    consume_memory(memory_size)
     consume_cpu(duration, cpu_load)
     print(
         f"Test completed. Consumed {memory_size} MB for {duration} seconds with CPU load factor {cpu_load}."