If the RAW was, say, '2+2', but never said '4', it wouldn't be mere 'conjecture', a mere 'assumption', that '4' is true as an unavoidable consequence of the RAW.
Here, the RAW that says the components must be provided or the spell fails results in the unavoidable conclusion that the spell effect absolutely cannot begin until the casting is complete, whether it uses those words or not.
That's not an unavoidable conclusion. For instance, in a normal IC engine, you must have a spark for it to start, but that spark still has to be there for the entire time it's running. You can't take a statement of 'X must be present for Y to happen. When Y happens, X will be fully comsumed' to mean 'X must be present and fully consumed for Y to happen.' Doesn't actually work that way. You're changing the logic of the statement to include conditions not actually present.
Zorku's sequence of events is not legal by the rules, simply because he has the spell effect beginning at a time when the components have not been, and may never be, provided. Therefore, RAW, the spell fails.
I disagree with your analysis. You can't start casting unless the components are present, and I don't recall Zorku saying otherwise.
There is no sequence of events that allows the overlap!
It's a bold statement, in that a single valid sequence would show that I was wrong at the top of my voice, and if such a sequence existed then I'd expect to see such a post appear sharpish, because who wouldn't want to burst so pompous a bubble? Which indicates how certain I must be.
Prick away!
You wish to cast a spell. The spell has v, s, and m component. In order to cast the spell, you must have an appropriate spell slot and the components present. At this point, you've reached the entirety of the necessary RAW to cast a spell. The only thing RAW wise you need to do now is use the cast a spell action (or ready). At the end of the cast a spell action, a few things will be true (absent counterspel): the effect will be in place (or done, if instantaneous), the components will be consumed (if consumed), and the spell slot will be consumed.
So, we can go with your interpretation, which says that the components and slot are consumed, and then the effect happens, and then the action is over. That's fine, but not required by the RAW. Or you can go with saying that you start with the components and slot present, then the effect happens, then the components and slot are consumed. That works by RAW, too, because there's no need to address actual timing in the spellcasting at all -- the only thing that can stop it is counterspell, and it works regardless of the order of events.
Personally, I like spellcasting that has the components woven into the delivery of the effect. Part of the S is point to where you want the fireball to explode, or which target the EB beam is aimed at. That works for me. And it still works if you're holding the spell with a readied action -- the release of the effect includes the final S or V or M pieces. If the spell is ruined due to damage and concentration in the meantime, the M component is still ruined because it's already charged with the magic and so burns away when the spell is ruined. The slot provides the charge, and so is spent regardless.
But, again, the fact that you've changed the logical statements of RAW to add timing not present and then insist that you're right because your other assumptions don't work unless you use your first assumption is really just an extended begging the question.