New ImageJ script runner

[petebankhead]

This replaces the ImageJ macro runner.

Both can be seen in this screenshot - the old on the left, the new on the right:

Improvements include:

• Syntax highlighting
• Autocomplete with built-in macro functions when pressing Ctrl+space
• Better control over which objects are sent for processing
• Better control over resolution, with 3 different methods to control the size of images sent to ImageJ (not just downsample)
• Better control over parallelization
• Option to select channels to send
• Built-in examples of potentially-useful macros
• Support for Roi groups, which map to QuPath classifications
• Scriptability; macros can be stored in the command history and run later through Groovy scripts
• Ability to run Groovy scripts instead of macros
  • These are different from Groovy scripts run through the script editor; they are specifically written to work on an ImagePlus obtained from ImageJ via a call to IJ.getImage()
  • New IJProcessing and IJFilters classes make it easier to process images obtained this way
• Properties are set in the ImagePlus based on the QuPath image type so it's possible to determine a sensible value of darkBackground or lightBackground in macros or scripts
• Better maintainability through the use of fxml and externalised strings
• Available by the 'ImageJ' toolbar button, not only the menus

This required a lot of changes, so there is a reasonable chance things have broken elsewhere... including with the regular script editor (which needed changed to make it possible to generate a script editor control elsewhere). But I think it remains functional...

Some limitations:

• It's not possible to control via a script which objects or Rois are sent to ImageJ with an OverlayOptions object
  • The previous macro runner didn't do this perfectly either - see ImageJ macro runner can't transfer both cell nuclei & boundaries #1522 )
• It's not currently possible to send back measurements with objects added to an overlay
• Only 'original' channels can be sent through the user interface, not color deconvolved channels (although scripting gives more options)
• The scripting language can't be set explicitly; for Groovy, ensure that IJ.getImage() or import [something] is included

[petebankhead]

This is the macro shown above, included as an example:

/*
 * ImageJ macro to apply an automated threshold to detect a single region.
 * You will need to return the active Roi to see the results in QuPath.
 */

// Define method (e.g. "Triangle", "Otsu"...)
method = "Otsu";

// Check if the image has a property specifying a dark background
// Override this by setting the value to true or false
if (Property.get("qupath.image.background")=="dark")
    darkBackground = true;
else
    darkBackground = false;

// Ensure 8-bit grayscale
resetMinAndMax();
run("8-bit");

// Create Roi from threshold
if (darkBackground)
    setAutoThreshold(method + " dark");
else
    setAutoThreshold(method);
run("Create Selection");

This effectively makes it possible to apply any of ImageJ's auto thresholding methods to any region of an image (or the entire image) - adapting for brightfield or fluorescence based on the image type.

The resolution and channel can be specified from within QuPath's UI when the region is being sent.