Populate dataset webclient

[sbesson]

Summary

Previously opened as #5691, but the metadata population logic has now been filtered out of OMERO.py into an independent component which can be released in isolation (see #5831).

This change cherry-picks the changes at the Webclient level for displaying Tables on P/D/I hierarchy. Minimally, when clicking on the Tables tab on an image, two additional queries are performed to load the bulk annotations at either the Project and/or at the Dataset level.

Testing

Two IDR studies have been imported under the public user of https://web-proxy.openmicroscopy.org/east-web/webclient/: the experimentA of idr0021 and the screenB of idr0008.

For each study, the bulk annotations have been populated using the annotation.csv files under the metadata repositories (idr0021 and idr0008) and omero-metadata:0.2.2

Tables should now load when expanding the corresponding tab in the right-hand panel for:

• Images under the idr0021 project
• Wells under the idr0008 screen (current behavior)

[sbesson]

Last two commits should contain a few additional fixes

• 3c0632c modifies the image query to use lowercase image-. This matches the current behavior of the metadata population tool e.g. see
  https://idr.openmicroscopy.org/webgateway/table/Project.datasetLinks.child.imageLinks.child/1884807/query/?query=image-1884807 vs https://idr.openmicroscopy.org/webgateway/table/Project.datasetLinks.child.imageLinks.child/1884807/query/?query=Image-1884807. The original behavior was implemented in tables data on images #4929 so I am unclear whether the Image-{{ manager.image.id }} case should also be supported
• 84eae09 should prevent an error message about undefined projectQuery variable in the developer tools when trying to open the Tables tab on a well.

[will-moore]

Working fine for Project and Screen on east-web as described.
Tried locally with a Table prepared for OMERO.parade and this isn't working as parade is using capitalised Image column (https://github.com/ome/omero-parade/blob/master/omero_parade/table_filters/data_providers.py#L101) instead of lowercase image column as expected.

Since we are using query="Well" elsewhere and WellColumns are called "Well", it is more consistent to use "Image" for ImageColumns.

[sbesson]

No personal objection to unifying towards Image- instead of image-. The consistency argument holds completely. Major concrete points to consider with regard to any long-term decisions:

• we are currently looking at omero-metadata, omero-parade and omerowebclient. Are there other consumers relying on this semantics?
• we should assume that current production instances like IDR already have data populated with an image column. What is the expectation for those? Are we forcing a systematic re-annotation in order to benefit from the Webclient and Parade integration? An intermediate might be to support both APIs in the meantime i.e. handle both image- and Image- queries but deprecate the former.
• independently of the path, we need to update our metadata population tool to comply with the format specification - see ImageColumn called  omero-metadata#9

[sbesson]

In order to test ome/omero-metadata#5 and in absence of concrete discussion and agreement, 572c319 implements the intermediate solution discussed in #5832 (comment) i.e. acknowledges the fact Project/Dataset tabular data currently exists under two forms and handle both variations.

[joshmoore]

@will-moore: is this good to go?

[will-moore]

Yes, this is working for me.
I guess the only question is how long we keep polling for ?query=image... and ?query=Image....

[joshmoore]

Thanks, @will-moore. I'll leave you to engage that conversation, creating cards as necessary.