OME-XML/TIFF export

[jamoore]

It should be possible to export first more metadata with a single image, to have those exports and their included metadata be configurable, to bulk export many images together (including screens or plates) and finally to perform these exports as a background process.

Usage

Possible API usage:

 exporter = session.createExporter()
 exporter.includeAnnotations();
 exporter.includeHcs();
 exporter.addImage(1L);
 export.generateTiff();

---

Notes

OME-XML (Josh, Jean-Marie, Chris, Andrew)

 * Requirements for improving OME-TIFF
    * Institute in Estonia
        * Putting 2000 images in one OME-XML
        * Lots of annotations in comments, long, boolean
    * Harvard Screening (Jay Copeland)
        * Export a single row / column / well group
    * EMBL Screening (Ruben Muñoz, Jan Ellenberg)
        * Not duplicating XML for each field, _plane_, etc.
    * Jean-Marie using OME-TIFF to communicate with ImageJ

 * OME-TIFF "lite", "basic"?
    * Just for display

 * Namespaces for annotations, etc.

 * Export
    * Where will we export from? (cF. cluster example)
    * Doing it in the client?
    * Andrew: Transporting one big TIFF and the client can piece it apart
    * Chris: exporter has to be modular
    * Jean-Marie: a danger is there are no clear strategies on namespace
    * Chris: Delete-like configuration, then choose the images
        * Josh: Instrument and a single plane?
        * Jean-Marie & Chris: Wait.

 * Using Metadata levels?
    * Josh: something similar in delete and in upcoming chgrp
    * Metadata levels are currently just 4 options: ALL, PIXELS-ONLY, ROI
    * Intentation as opposed to implementation
        * Not as flexible as delete
        * That flexibility is probably overkill for export
    * Then technically do the export
        * TIFF per file
        * All ZIP'ed
    * Andrew: Metadata-only files?
        * OME-TIFFs as the companion file with minimal spec
        * Pseudo-schema on top of 2010-06 in OME-TIFF (only <OME uuid=""/>)
        * Lots of coding work

 * Steps
    * Delete-style options
    * Data-layout options (zip, single plane, etc.)
    * Model: how to split the files. (release before server?)
        * Andrew: publish but don't release (PREVIEW)
    * SA best practices
        * Have to show it in insight/web!
    * Round-tripping
        * LSIDs working
        * Have a graph which was written in the original import (utility of that?)
    * Full OmeroReader implementation (all 400 methods)
        * a couple of individuals; reverse of the Store
    * Timescale: something by mid- to late-January

[jmoore]

Moving to Unscheduled since we are approaching the 4.3 freeze.