First attempt at offset controller driver.

Do you think it makes sense to write a simulator and testcases for this driver? If so, what would you test for? We don't get any values back from the hardware if I understand it correctly, so there's not much we can do.

This PR contains all the acquisition panel including the zstack capability. All the code related to the zstack acquisition is only in the second commit (3951cfa).

Megafield, overview, and livestream acquisitions need different settings of the scanner. The new function sets these values based on keyword arguments which can be passed through a dictionary.

• bugfix ROC: couldn't be selected at high zoom level
• only enable calibration buttons for scinitallators selected in chamber tab
• use storage path from asm

Fixes for FLIM. Fixes

• SPARC, SECOM, Delphi are working. Further ongoing testing is still ongoing.
• Rep control moved to stream controller
• Dwell time VA does not affect ROA overlay
• Dead object errors fixed
• FLIM stream merge ratio fixed
• Overview working
• Incompatible tool modes disabled on some displays

Some issues still persist:

• It is possible to set a pixel size and refresh it, but somehow the ROA recalculation still hampers the ease of use
• ROI selected by default in SPARC

…the dwell time

Update the estimated acquisition time when the dwell time on the ebeam scanner is adjusted. Also improve the estimated acquisition time. Now it used an fixed estimate per tile based on a dwell time of 1us. Now calculate the time per tile based on the number of pixels per tile plus some fixed overhead for stage movements etc.

We pause the stream in between each acquired image, this already blanks the beam. Forcing blanking and subsequent unblanking of the beam caused it to blanked during acquisition sometimes. In the fastem configuration set the blanking to False for live view and overview imaging, so it's unblanked during acquisition.

Include skimage's peak_local_max function as it allows more options for filtering spots: intensity (absolute and relative), distance, etc. And it's faster as well.

_get_intersected_field_indices() Assumes the y-axis as being positive in a downward direction. To convert this to the real world definition the acquired index array needs to be flipped over the x-axis.

Running /home/testing/development/odemis/src/odemis/gui/test/comp_legend_test.py:
FAIL comp_legend_test.py::LegendTestCase::test_legend

remove blocking function

The checks on the filter axis were confused. In particular, it assumed that the current filter position was not the second position randomly picked... which was sometimes incorrect. Also, the axis takes typically 1.1s to move. So 1s is too short to wait.

Resolves wxPython warning (SWM-73). The warnings were from the tests in odemis/src/odemeis/gui/test

Solution:- Install PyPubSub package which is actually what wx.lib.pubsub was based on Then replace from wx.lib.pubsub import pub with from pubsub import pub

Warning Message:- /usr/lib/python3/dist-packages/wx/lib/pubsub/__init__.py:23 /usr/lib/python3/dist-packages/wx/lib/pubsub/__init__.py:23: wxPyDeprecationWarning: wx.lib.pubsub has been deprecated, plese migrate your code to use pypubsub, available on PyPI. warnings.warn('wx.lib.pubsub has been deprecated, plese migrate your '

The simulator didn't properly handle stopping. If a measurement was stopped and just after a new one started, the first one would somehow "continue"... in a very weird way, as all the rest of the code assumes that only one measurement is running at a time. It could even happen if starting a new measurement just after one finished, as happened in the test case test_acquisition().

=> Stop using a RepeatingTimer, and introduce a dedicated class to simulate measurement, using a command queue and FSM. It might be slightly over-kill but works nicely.