Hude’s Oxford Classical Text has συμβληθηναι with no sigma (and lists it as the last word of 3.125.2, incidentally), and the app. crit. doesn’t list your form as an alternate. συμβληθηναι, of course, is the aorist passive infinitive of συνβαλλω. Which text are you using?
The normal outcome of two dental consonants right next to each other is for the first one to become a sigma. For example, the aorist passive of ᾄδω is ᾔσθην, with the delta becoming sigma before the theta. This sigma was generalized out to a bunch of verbs where there’s no historical reason for it to be. We will see it sometimes in that popular paradigm example παύω, i.e. ἐπαύσθην. This seems to have happened here.
Smyth 489 has a long discussion. He mentions that the sigma becomes increasingly common in later manuscripts.
Hmmm…how sure are you about this “intrusive analogical sigma”? Among aorist passive forms of [size=150]βάλλω[/size] I can find no evidence for any such formation. Nor do I find it attested anywhere, at least not within the Perseus collection.
The edition which I checked was the very same text (the OCT that is, not the transcript!) - and the errant sigma is not there. Given that the comparision is a transcript of this text, I believe the typo theory is the correct one.