[REVIEW] Fix brute force KNN distance metric issue

[viclafargue]

The brute force KNN default distance metric switched from L2SqrtExpanded (or equivalent with FAISS metrics) to L2Unexpanded recently. I believe that it was a voluntary move to speedup calculations. However some calls to the raft::spatial::knn::brute_force_knn function in the cuML codebase required specifying a non-default metric. This is the case of distances that are not kept internally by an estimator but are instead provided to the end-user as results. This is also the case of features that specify a specific distance metric for their calculations. Please tell if you notice some other places where a non-default distance metric should have been specified.

cc @MatthiasKohl

[MatthiasKohl]

Concerning MNMG KNN, the change looks good.
I can see that tsne and umap (as well as the tests for umap) also use this raft::spatial::knn::brute_force_knn primitive. I'm not sure whether for those algos the metric matters, but maybe someone should double-check?

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (branch-0.20@ef78253). Click here to learn what that means.
The diff coverage is n/a.

@@              Coverage Diff               @@
##             branch-0.20    #3755   +/-   ##
==============================================
  Coverage               ?   85.96%           
==============================================
  Files                  ?      225           
  Lines                  ?    16986           
  Branches               ?        0           
==============================================
  Hits                   ?    14602           
  Misses                 ?     2384           
  Partials               ?        0           

Flag	Coverage Δ
dask	48.96% <0.00%> (?)
non-dask	77.80% <0.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ef78253...afdc477. Read the comment docs.

[viclafargue]

@cjnolet and @lowener would be nice if you could take a quick look at this PR. Do you know if I forgot any necessary modifications here?

[lowener]

LGTM

[divyegala]

LGTM