I really question whether anyone buys into a "St Rob" narrative. Some of the polls suggesting significant support for Thug were ludicrous, notably Forum had him topping out the 18-34 age group.
As for Mainstreet, they were the only ones to release methodology and sampling weights. Otherwise they used the same technique as Forum, yet got a wildly different result for Thug. It is simply not credible that the heretofore unpopular Doug Ford is getting some kind of "bump" from his cynical entry into the election.
Anyway, we have seen ample evidence in recent years that polling has become increasingly unreliable. In a municipal election, the sources of sampling error and bias may be even worse, and the wild within strata swings suggest this.
Forum also has been known incredibly biased questions - asking for preference between LRT and subway in Scarb on the false premise that they have equal costs.