[FEA] in SVC, support_ is updated in fit() when dataset contains two classes, but is not updated when fitting multiclass data

[NV-jpt]

Describe the bug When fitting SVC to certain data the .support_ attribute does not get updated

Steps/Code to reproduce bug In this gist, you can see that cuML's SVC does not update the support_ attribute after fitting to this multi-class dataset.

https://gist.github.com/NV-jpt/a6243373c784f2efa8f708e5ed691940

Expected behavior I expected cuML's SVC to behave like that of sklearn's. In the above gist, you can see the behavior of the sklearn classifier.

Environment details (please complete the following information):

• Environment location: Bare-metal
• Linux Distro/Architecture: [Ubuntu 18.04 amd64]
• GPU Model/Driver: [V100 and driver 460.32]
• CUDA: [11.2]
• Method of cuDF & cuML install: from source cmake version 3.10.2, gcc (Ubuntu 9.4.0-1ubuntu1~18.04) 9.4.0, commit 277b91e961b5861b51726afab42c27b8bb0fe60d

Additional context This bug is preventing the integration of cuML into imbalanced-learn's SVMSMOTE resampler in the case of multi-class data.

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[github-actions[bot]]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.