We need proper counterfactuals here, cases where a practical use of math counterfactually would not have been possible without previous development as pure math.

I think a decent argument could be made that Einstein would have been unlikely to have been able on his own to work out the necessary math he used in special and general relativity. On the other hand, this is much more of a severe issue for gen relativity, and it isn't implausible that once he had constructed the basic theory others would have listened to him enough to work out the underlying math.