Studio-NDTiffAdapter: Simplifies function getImagesMatching()

[nicost]

so that it does no longer go through the complete dataset. This may lead to problems in certain cases, but at least makes virtual datasets readable again.
Closes #1839

[marktsuchida]

Just to check, is the intent not to change what getImagesMatching() returns, but to change how it computes the matching set?

This may lead to problems in certain cases

Which cases?

[nicost]

I believe it now only returns an exact match, so no longer the super set (i.e., if one only asks for channel == 1, it should return all images with channel == 1). I do not know how to implement that behavior with the NDTiffStorage back-end.

[henrypinkard]

(Edit)

I thnink the problem is that a function that reads the metadata but not the image had never been added to the NDTiff API. I just added one:
micro-manager/NDStorage#142

[henrypinkard]

This is the line that I think would need to change

[nicost]

This is the line that I think would need to change

To what? I do not really understand this code.

[henrypinkard]

from storage_.getImage to storage_.getImageMetadata (so that it doesnt read pixels when it doesnt need them)

and then change the addEssentialImageMetadata to do whatever it is the studio needs

[nicost]

What I would need from the storage class is a method that tells which axes are available and which images (with axes coordinates) are in the dataset. It it unclear from the NDTiff Storage API is such a method exits, and documentation such as:

 /**
   * Get the metadata for a single image
   *
   * @param axes
   * @return
   */
  public JSONObject getImageMetadata(HashMap<String, Object> axes);

is not very helpful. I will merge this PR since it at least improves the situation, but welcome work on making this actually work right.