It's very useful to have some kind of robustness reference. You can refer to your current data that they are robust. Note that the statistical problem does not affect your main finding. This is verification that your work is good without new math. It helps to keep your paper focused on your original data. The only opener I have found so far that works is: "To rebuttal reviewer's response on [some issue], we remind that... the model is robust, as depicted by...." This line demonstrates that you appreciate their opinion. That also explains why your approach is correct. This way, you can let your accomplishments shine. It deals with the criticism professionally.
Thanks for bringing up heteroscedasticity. We actually checked that. In our recent IEEE paper, we just showed the residual plots and diagnostics. They were clean enough, so we didn't add more tests. I've found that explaining what you already did is usually enough to satisfy reviewers without creating extra work for yourself.
When a reviewer calls out stats like heteroscedasticity, I just start with, "Thanks for pointing that out." Then I explain our approach, maybe mentioning our statistician already confirmed the variance checks were fine. In my experience, it's better to be open about what you did and show it follows the field's standards. Don't promise to run new analyses unless you absolutely have to.