perf: use batch loading for BasicUserDetails

[ACLay]

Description

Expands the user batch loading from #469 to apply to BasicUserDetails as well, since the basic variant is quite widely used, and thus causes a lot of calls to the slow external user service at STFC. By combining multiple external calls into one we'll have to pay the time penalty of making network calls fewer times, and when that's ~100ms for each call in our prod environment, it'll add up to quite a bit of user-data-heavy requests.

Motivation and Context

How Has This Been Tested

I made 3 local page loads of access panel 11's assignments, recording the time taken by the main 3 calls when starting with an empty users cache. Due to react's dev setup making calls twice, this results in 6 calls for each. For those runs I also tracked the number of UOWS calls made. All times are in milliseconds

Method	Call time (current client)						Call time (with batching)						Average time saved	Percentage improvement
getFap	87	89	83	83	104	125	66	68	54	55	73	77	29.666	31%
getFapMembers	716	735	704	721	674	675	446	564	589	577	622	642	130.833	18%
getFapProposals	711	741	702	725	645	674	538	562	593	581	612	644	111.333	16 %

Attempt	UOWS calls - no batching	UOWS calls - batching
1	71	38
2	86	48
3	72	49

Fixes

Closes UserOfficeProject/issue-tracker#1197

Changes

Depends on

UserOfficeProject/mock-uows#19

While the test suite as a whole is passing here, some tests in isolation would have failed without this update to the mock uows, which has already merged. The initial passing state was I believe due to requests for individual users being made first in other tests and then cached in the backend. Without that, any test that first requests multiple users with batch loading would have failed as the mock server would only give one user's details.

Tests included/Docs Updated?

• I have added tests to cover my changes.
• All relevant doc has been updated

[ACLay]

This gets rid of our old windows container debug profile and replaces it with the newer one that our debug instructions on confluence advise putting here. Once this merges in, that section will need to be updated https://user-office-project.atlassian.net/wiki/spaces/UOP/pages/19136712/STFC+development+workflow#Debugging-the-backend

[TCMeldrum]

Looks good to me. Out of interest if the number of calls to the UOWS is cut down by a half does that mean only 2 people are being batch loaded at a time?

[ACLay]

Seems like it. I guess the batching might be done at a more local level within each query, like if we're loading proposals on the panel one by one, it's probably not the fastest to batch up and only request the reviewers at the very end. Plus there's likely going to be a lot of repeats in the reviewers requested in the panel, so a lot of requests are already be dealt with and reduced in the initial unbatched case by the caching.