[quant] Explictly set default quantized engine instead of relying on the order of supported_qengines (#89804)
Summary:
Fixes: https://github.com/pytorch/pytorch/issues/86404
Test Plan:
ossci + sandcastle
Reviewers:
Subscribers:
Tasks:
Tags:
Differential Revision: [D41635738](https://our.internmc.facebook.com/intern/diff/D41635738)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/89804
Approved by: https://github.com/andrewor14
