Not really. Suppose that the first is stated by the GM, the player makes a Resist Passion roll, and fails, and then the GM state the second. How did this situation suddenly change from "test" to "manipulation"?
It didn't. The invokation and use of a mechanic (the Resist Passion roll) kept it at 'test', which was then failed, thus giving the GM the right to narrate the results of said failure. All is good.
But the second statement
without any mechanics involved isn't a test, it's a manipulation.
Or to give a different example. The GM has described the dungeon corridor that the PCs are standing in. The player says I walk down the left-hand path, inspecting the ceiling as I go. The GM responds, OK, after about 10' you find yourself falling - that bit of floor was an illusion! Is that "test" or "manipulation"?
It's magic, which allows a certain amount of bypassing the normal rules. Same idea, in a way, as how a charm spell allows a DM to make a PC do/feel things he might otherwise not.
However, let's say it wasn't an illusion but just a simple pit. The player has stated the PC is specifically looking at the ceiling, so the GM just deciding that the PC doesn't see the pit coming* might be fair game; though I think most would give some sort of perception roll in any case and have that one roll kind of serve two purposes: a) did the PC notice anyhting odd about the ceiling and b) did the PC happen to notice the pit ahead.
* - one has to ask why the rest of the party aren't warning this poor sot to watch where he's putting his feet.
I wrote the OP, so I can condidently say that you are wrong about this. The OP says nothing in particular about what the mechanics and system conventions might be around establishing true descriptions of PC actions - for instance, what resources might need to be spent in order to be permitted to make a description true. It deliberately and expressly makes the range of possibilities a matter of discussion!
Actually, that it says nothing about the mechanics and system conventions being used says to me quite specifically that there are none being used at all (otherwise they'd have been mentioned, hm?) and thus it's an example of a player dictating an NPC's reaction.
I think you may have missed the point of the OP. I described an action - I wink at the maiden, melting her heart - in the course of inviting discussion about how these descriptions of actions might be made true of the fiction. The OP canvsasses decision-making and checks - for D&D players, this at least roughly corresponds to the difference between spell-casting and thief abilities.
I don't know why you would equate a player decision-amking ability with bypassing game mechanics.
But that's just the point: you didn't just describe an action. You described an action (winking at the maiden) and its result (melting her heart) all in one. The action is fine, but describing the result without reference to either it being an attempt only or to any system mechanics or conventions is where the problems arise: it reads as if game mechancs ARE being bypassed by player fiat - which is why I turned the example around to make it GM fiat so you and others could see the problem for what it was.
The whole point of the OP was that simply saying The players decide what their PCs do isn't a useful description of any RPG, given that I wink at the maiden, melting her heart is a true description of what a PC does, but isn't something that a player normally has the unfettered power to make true in a RPG.
Actually, no it isn't.
'I wink at the maiden' is a true description of what a PC does. 'Melting her heart' is merely a description of, one must assume, the PC's goal in doing it; but without the co-operation of the maiden there's no implied guarantee that this goal will be achieved...except by use of either game mechanics (which aren't mentioned) or GM fiat (which, as mechancs aren't mentioned, becomes the default). It's merely an attempt, and thus should be phrased as such.