Yes, quite rare, even in Attic Greek.
Since authors actually write to communicate, and want their readers to understand, context usually provides sufficient clarification. They tend to avoid "weird" constructions (but not always, calling as witness the book of Revelation). Yes, there are contexts in which true ambiguity occurs, but those are relatively rare. What happens too often in biblical interpretation is theological motivation to read the text a particular way.I read the OP's question the same and I have to balk at the notion about asking whether an interpretation of a particular text is "grammatically possible" because the question is asking about what the text could mean in isolation, but the text does not occur in isolation but in context. As a general matter, lots of interpretations are "grammatical possible," even weird ones, as long as there is contextual support for it.
I'm not sure I agree, which is why I cited BDAG above to demonstrate the normal usage of ἐτοιμάζω with the dative. I think knowing the first century background is helpful, but not absolutely necessary.In this case here, the contextual support for disambiguating the dative is almost entirely theological: how first-century Jews understood who prepared the eternal fire. The text is not teaching new concepts about the eternal fire but presupposing them. Answering the interpretative question as a result cannot avoid the (first-century) theology.
Now, being a Latin language sort of person, I wondered how Jerome rendered this:
qui paratus est diabolo et angelis eius...
Now, agency in Latin is practically always expressed by the preposition ā/ab + ablative. Like Greek, there is a dative of agent, but also like Greek it's restricted to one context, when used with the passive periphrastic (don't ask). Diabolo and angelis could be either ablative or dative, but paro (paratus) often takes a dative just like its Greek counterpart, and if Jerome had intended agency, he would have written ā diabolō et angelīs. Jerome therefore understands the text as using a dative of reference with the verb, as has just about everyone else in the history of interpretation.