I meant "can have no subject" as in "it is possible to have no subject", not "it is not possible to have a subject". That is, I don't think it's quite right to say that it's possible for a {-meH} verb modifying a noun to have no subject at all.

Unless I've missed something, the only situation where we know that a verb is considered to have no subject and to not be in any person is when it's used with {-ghach}. Given the framework of the language as we currently know it, it makes more sense to me to interpret {ghojmeH taj} as having an impersonal third-person verb, rather than a subjectless quasi-infinitive.

We know that Klingon has impersonal third-person verbs that are still conceived of as having unspecified subjects: {SIS}, {taH pagh taHbe'.}

We know that some noun-modifying {-meH} verbs have explicit subjects: {jIpaSqu'mo' narghpu' qaSuchmeH 'eb.}

We know that in other instances where English or other languages usually use infinitives that Klingon still requires a subject, explicit or otherwise, like in sentences with {'e'} or {neH}. E.g. "I want to drink" is translated with an explicit subject for "drink": {jItlhutlh vIneH.}

It's possible that some {-meH} verbs modifying nouns are another exception to the "verbs have subjects and persons" pattern like {-ghach}, but Maltz hasn't said so one way or the other. Until he does, I don't see a reason to make an exception to this pattern just for some uses of {-meH}, when the existing pattern can handle those uses just fine as impersonal third-person verbs.