entangle.AnalogOutput: fix window mechanism

fix 2 problems:
- in case the window is smaller than the pollinterval, isAtTarget might be False
  for up to the timeout delay
- in case the history is shorter than the window, and the ramp is fast enough to miss
  any points during ramp, isAtTarget is True before the window time is reached.
  This happens because the history is reset on write_target

+ do not wait when target is not changed (by more than precision)

tested!

Change-Id: Ia4ff4378fe91fa93be50168b2883a20b49ebfb6a
Reviewed-on: https://forge.frm2.tum.de/review/c/sine2020/secop/playground/+/27159
Reviewed-by: Markus Zolliker <markus.zolliker@psi.ch>
Reviewed-by: Enrico Faulhaber <enrico.faulhaber@frm2.tum.de>
Tested-by: Jenkins Automated Tests <pedersen+jenkins@frm2.tum.de>

secop_mlz/entangle.py

@@ -492,8 +492,11 @@ class AnalogOutput(PyTangoDevice, Drivable):
         hist = self._history[:]
         window_start = currenttime() - self.window
         hist_in_window = [v for (t, v) in hist if t >= window_start]
+        if len(hist) == len(hist_in_window):
+            return False  # no data point before window
         if not hist_in_window:
-            return False  # no relevant history -> no knowledge
+            # window is too small -> use last point only
+            hist_in_window = [self.value]
 
         max_in_hist = max(hist_in_window)
         min_in_hist = min(hist_in_window)
@@ -574,7 +577,9 @@ class AnalogOutput(PyTangoDevice, Drivable):
         if not self.timeout:
             self._timeout = None
         self._moving = True
-        self._history = []  # clear history
+        # do not clear the history here:
+        #    - if the target is not changed by more than precision, there is no need to wait
+        # self._history = []
         self.read_status()  # poll our status to keep it updated
         return self.read_target()