So, this seems to be a question of philosophy of language and epistemology since it is predicated about notions of the analytic-synthetic divide, questions regarding syntax and semantics, and is examining propositions in light of their linguistic modality. To wit:
- If imperatives don't have subjects, then how can we describe them as "analytically justified" if analysis is supposed to be extraction of predicates from subjects? Is the grammatical "trick" (in American English at least) of an "understood" subject such as "You" sufficient?
- He does this not just for assertoric forms of the CI, but also the purely prescriptive form that starts out, "Act only..." instead of, "You ought to act only..."
The differences among the assertoric nature, apodicticity, and normativity of propositions are properties of the question of the self-evident nature of the proposition and indications of propositional attitude. Thus, a simplification of your question, that is to say a restatement of the content of your statement, might be thus:
Is any statement such as "Act only..." analytically justified given the fact that the literal statement does not contain a subject, and does that differ whether the proposition is an imperative or a normative statement?
analytic proposition: a proposition whose predicate concept is contained in its subject concept
synthetic proposition: a proposition whose predicate concept is not contained in its subject concept but related
The answer to this question of the sufficiency of grammatical construction as a basis for the justification of the classification of propositions might be answered differently in Kant's day than in modern philosophy, as Kant predates linguistics, psychologism, and cognitive science (presuming your epistemic attitude isn't hostlie towards). In addition, Quine's attack on the analytic-synthetic distinction is also relevant, to your question. (SEP: The Analytic/Synthetic Distinction)
Note that criteria listed above cite the propositional content, and not the grammatical form. Kant was more interested, inferentially, in the concept of a proposition being self-referential in some regard to distinguish tautological propositions from non-tautological ones as a precursor to moving onto establishing a dichotomy related to questions of a prioricity and a posteriority of propositions. One is forced to explore questions of semantic identity, definition, equivalence in rational discourse, and then further to deal with how these representations match the state of affairs in the world that one must learn.
Hence, it seems that implied subject is NOT a reason to discount the analyticity of a proposition, and the question of the elliptical construction might have merit only to those object to semantic realism or reject transcendent meaning. Philosophers such as Quine, Davidson, and Derrida deny such language-transcendent meaning exists, and are much more insistent that syntactical expression is the best path to meaning. But Quine, in particular, rejects the analytic-synthetic division as meaningful, so this question would be meaningless to him!