How I answer review requests these days

First answer:
I need a full week to look at the paper and then I can tell you if I feel qualified to review and how long it will take.
Usually the Associate Editor will gently let me know that it is not the policy of the journal etc ... Which triggers my second answer

Second answer:
I do the review. For free. And I want to do it seriously. So I will not let obnoxious journal policies decide for me.

Some AEs would then ask me why I actually need a full week to decide if I can review a paper and why I should impose the duration of the review process. So there I go.

Third answer:
If the paper is about maths, I need to check the level of these maths. As a reviewer I will have to check the theorems. One by one. It takes time. And if I don't feel I have the necessary knowledge I simply shouldn't do the review. Now if the paper is experimental, I should check the experiments. Do the authors provide the code and data ? If not, I should implement the algorithms. Yes, all of them. It can literally take months, but this is what reviewing is all about. It's about making sure that the paper is correct, presents facts, in a verifiable way.

Take home message:

AE: I do the job, you let me do it my way. Or you find another reviewer
Authors: make sure the reviewers get your code, your data. Document them. It will only make it easier to asses your work. Everybody wins,