#69 - Fix an issue with the callback iteration timestamp drifting

The callback iteration slowly drifts a few milliseconds per call
This is due to inaccuracies of time.sleep and some overhead on the `_run_callback` method

This fix the issue by using sched.enterabs instead of sched.enter.
This time to enter is always calculated based on the start timestamp of the callback

There will still be some milliseconds variation from execution to execution, but it will never drift.

dynatrace_extension/sdk/callback.py

@@ -45,7 +45,7 @@ def __init__(
         self.ok_count = 0  # counter per interval = 1 min by default
         self.timeouts_count = 0  # counter per interval = 1 min by default
         self.exception_count = 0  # counter per interval = 1 min by default
-        self.iterations = 0 # how many times we ran the callback iterator for this callback
+        self.iterations = 0  # how many times we ran the callback iterator for this callback
 
     def get_current_time_with_cluster_diff(self):
         return datetime.now() + timedelta(milliseconds=self.cluster_time_diff)
@@ -140,5 +140,6 @@ def get_next_execution_timestamp(self) -> float:
         This is done using execution total, the interval and the start timestamp
         :return: datetime
         """
-        return (self.start_timestamp + timedelta(seconds=self.interval.total_seconds() * (self.iterations or 1))).timestamp()
-
+        return (
+            self.start_timestamp + timedelta(seconds=self.interval.total_seconds() * (self.iterations or 1))
+        ).timestamp()