Use active_user.id instead of object in autocomplete REST data

[will-moore]

Noticed while comparing URLs with those from IDR gallery.

Previously URLs to load auto-complete data looked like e.g.

http://localhost:4080/mapr/api/autocomplete/gene/?... ..experimenter_id%5Bid%5D=-1&experimenter_id%5BfullName%5D=

since the whole WEBCLIENT.active_user object was being serialised to the query. With this fix, it should now look like experimenter_id=-1 (inspect XHR requests in devtools).

To test:

• Go to mapr, e.g. https://merge-ci.openmicroscopy.org/web/mapr/anyvalue/
• Open devtools in your browser, click on Network tab and XHR.
• Start typing in the mapr search input (above the tree). It doesn't matter too much what you type but it will look for 'any' matching Values from map annotations.
• You should see requests in the network tab that have experimenter_id=123 in them, e.g. https://merge-ci.openmicroscopy.org/web/mapr/api/autocomplete/anyvalue/?value=in&query=true&case_sensitive=false&experimenter_id=454&group=454&_=1568968925363

[will-moore]

NB: when this is deployed on IDR, the autocomplete URLs will be uncached. But I don't know if any cache population is needed (lots of combinations possible for auto-complete). cc @sbesson @manics.

[mtbc]

I found that I seemed to get something useful if I logged in and added some map annotations then searched for those values.

As user-1 on merge-ci the request URL does include experimenter_id=452 so I think this test passes? That is indeed the user's ID.

Not quite sure what to make of what happens next. I pick a valid search result from the list of values but I don't then see the image in the tree, it just twitches a bit then goes back to how it was. Perhaps that's expected and/or outwith this PR.

[mtbc]

(Ah, if I put the KVP on an image that's in the dataset then I at least get a dataset count even if I still can't navigate to it. Again, maybe outwith this PR.)

[sbesson]

Thanks @will-moore. In the case of a cached deployment like IDR, my suspicion is that the old autocomplete cache will become stale and new URLs with the fixed experiment parameter will be generated.

Given the positive review, merging and releasing as 0.3.2. We will investigate the impact into test71 and come back to this repository if necessary.