Is poor English grammar and writing style overlooked in otherwise strong math paper submissions?
Short answer: no
Longer answer: The purpose of the article is to communicate something. If poor grammar and style gets in the way of that goal, then that is something that needs to be addressed before the article can be published. If the paper is indeed strong, then that will typically mean you will get a "revise and resubmit" rather than an outright "reject".
Sometimes (as a referee for a submitted paper) I have provided detailed grammar corrections1. Other times I have just said: "The English is unacceptable, get someone more expert in English to correct it for you." I do seem to recall in one case the editor told me that the journal staff will correct the English, so I should not worry about it, instead only worry about the technical content.
1 éventuel does not mean eventually
Are you asking about the "should" or the "is"? It depends on the referee and the editors. I have seen papers published in top venues that would have seriously profited from 5 minutes of proofreading (sometimes even from an automated spellcheck). But I have also seen referees point out rather subtle linguistic errors carefully and in detail (and have done so myself a few times). There is probably no shortage of referees who are willing to rubber-stamp a paper written by a celebrity no matter whether any of it makes sense. I also wouldn't be surprised if an editor feels pressured to publish a paper by a VIP even despite the objections of referees.
There is also a huge distinction between the kind of errors that merely reveal the author to be a foreign speaker, and the kind that make reading difficult. On occasion, a language barrier completely prevents the author from readably communicating a proof; while I cannot blame the author for the "mistake" of being born in the wrong country, I don't hesitate to send back such a paper for revision, after giving as many suggestions as I can (and as concrete as I can) for how the writing could be improved and what pieces of syntax they seem to be missing. Things like this should not lead to rejection of papers, unless the author fails to clarify their work through several revisions; but the paper should not get accepted until this work has been done.