It is true that very frequent words (whether nouns, prepositions, conjunctions, or whatever part of speech) can undergo one-off changes, usually involving dropp

It is true that very frequent words (whether nouns, prepositions, conjunctions, or whatever part of speech) can undergo one-off changes, usually involving dropping syllables. But the phonetic details don't make sense here as a reduction: with is arguably harder to pronounce than mid. You're basically arguing for a rare and poorly understood kind of lexical change with no evidence to support it, against a well-understood and very common kind of change with multiple different strands of converging evidence. It's not impossible that the phonological similarity of "mid" and "with" made it easier for the meanings to coalesce in "with". But it is certain, at least as certain as you can get in historical linguistics, that these were originally two prepositions with independent origins.