File streaming and concurrent acquisition support

[jacopoabramo]

I did some refactor on the RecordingManager and also the DetectorManager.

What I did was:

• use the napari threading framework to spawn multiple QRunnable to stream data from a detector to a file;
  • one thread is spawned for each detector currently flagged forAcquisition;
• add streaming support for OME-TIFF and HDF5 files support;
• change the getChunk method in the DetectorManager class to return a tuple of two numpy arrays:
  • one for the actual data chunk to be written;
  • one for a list of frame IDs used for checking possible frame losses during long acquisitions;
• add a synchronization mechanism to update the GUI whenever a new recording is done similarly to what was already being implemented.

Not everything is currently supported:

• time lapses are not supported yet;
• the synchronization when using multiple cameras is a bit buggy;
• zarr streaming is still missing;
• data chunks are continously collected by a direct call to the getChunk function of the DetectorManager class; this is inefficient when one has (like I have) an instrument which aims for going at very high acquisition speed (i.e. hundreds of microseconds of exposure time, even lower).

What I would like to have:

• a thread-safe memory buffer which allows a camera to write on a specific RAM section continously, while the streaming object continously write it. If the streaming object is fast enough there should be no data losses.
• for reduced data storage requirements, support for lossy/lossless compression algorithms would be nice.

EDIT: fixed links