Meanwhile back at the project that started this all off.
Again, about names, 'nimi ijo' was once sorta agreed to for nouns and 'lawa' doesn't add much to it. 'monsi' is only significant if you know that tp is NA, which hasn't been mentioned though probably displayed. And, of course, words behind won't necessarily be adjectives/adverbs (nor adjectives/adverbs be words behind). Modifieres (adv/adj) are often 'nimi ante, though I think that was the most controversial naming.
the 'ijo x' -> 'ijo' derivation of nouns is like part of one good theory about building NPs, but the examples are cass of it, since 'suno' is a noun to begin with (so 'ijo suno' is derivative) and 'ijo moku' is a thing that eats, not food at all ('moku' is food because 'moku' means "eat" and food is the genus of all proper DOs for that verb).
Prepositions are, not surprisingly, 'nimi sinpin' What are they inside of? They can, of course, have adverbs or adjective, certainly 'ala' possibly others. The cases of preps dba verbs are odd. The first is simply wrong, since, when acting as a verb, a prepsoition has all the ptentials of other verbs while keeping the tight object construction (complement immediately after the prep with all its additions) . But the fact that a direct object to the transitivized prep has to go after its complement is right. The shifting of the complement ot a repeated version of the prep at the end is also a common stylistic variant, especially when the decorations get too fulsome: 'mi tawa noka wawa tomo' -> 'mi tawa noka wawa tawa tomo'.
Verbs were called "nimi pali". I am not clear why verbs are important and other words not, although "verb" does mean simply "word" originally and "noun" meant the special sort, names, while the others were just atached to them. I guess I would have (were I to do this division, which I don't see areason to) put nouns and verbs both into the suli class, since you can't have a full sentence without one of each, at least.
Rather to my surprise, I find your explanation of the source of 'mi wile e ni' basically correct.
There are at least four kinds of questions in tp, so calling 'seme' the question word is a bit misleading, even though it is almost the only word that makes questions all by itself. 'anu' needs a question mark or intonation to make a question, and the intonation alone will work with any sentence. The other two kinds of questions are reasonably accurately described but 'pi nimi pi 'lon' anu 'ala'' or something like that (I have no hope tht you will get the point of that). And, of course, prepositions dba verbs fit perfectly well into the basic Y/N form: 'sina kama ala kama musi?' and the second example of 'tawa' is ambiguous and unnecessarily so.
The possibility of subordinate clauses is in the air but not much more than a thought at the moment. I will assume you are opposed to proceeding with the thought. Neither example you give is what is being thought about, however. The first would be (as things go so far) 'mi sona e jan pi li moku e soweli' (though your particular case would not be a problem) and the second seems to be a different beast all together, not a restrictive relative clause but indirect discourse, which hopefully will not get special treatment with the perils of subjunctives and the rest lurking in the ofings for SAE folk.
Your comment on 'lon' and 'la' is correct, but incomplete. The 'lon' is lost with all (?) cases of 'lon tenpo' in the terminal PP position, but not certainly with other core words (some seem to, some seem not to and there doesn't seem to be an obvious pattern). I think making your suggestion the rule for such shifts (which tend, in fact, to be taken as adding a 'lon' when moved from 'la' to the end) would simplify life and not create problems (so far as I can see now).