Is More Flow Better, Is The Smallest Intake Port That Flows The Most The Best

LOL…you can only get some much stroke. Then what?

The math is:

P.
L.
A.
N.

Exactly.
Instantaneous torque would be defined by the crank throw along with rod angularity, bore, and pressure within the cylinder. There's only so much pressure which can be generated, only so much bore a block allows (controlled by bore spacing), and only so much throw a crank can have. Short rods make long strokes a diminishing return as well.
But RPM, that's only limited by momentum. A good cam lobe and proper springs with stiff pushrods and as little weight possible at the valve can up the RPM potential (if the head can keep up) pretty quickly. High rocker ratios help too, but between the piss-poor lifter angle and an unwillingness with most to mill-off the rocker stands, there's only so high a ratio we can fit. Plus, as ratio goes up, so does the force at the pushrod which makes the attachment hardware even more critical. During overlap, two rockers with springs compressed at 600+ lbs starts to stress things and send them on a walk.

3.58 to a 4" stroke is an 11.7% increase in stroke/torque (not really, but for argument's sake).
This is equivalent to an increase from 6k to 6704 rpm, or from 7500 to 8380 rpm.

This tells us a few things: as rpms increase, it takes bigger increases to net the same gain. Going from 9k to 10k (11% increase in revs) won't net the same as going from 5k to 6k (20% increase in revs). The same is true for an increase in stroke without an increase in rod length. But long rods tend to push decks higher, make intake tracts longer, and push carbs higher - which all have limits as to what's reasonable as well. People are afraid of RPM because they don't like to pay for balancing (pre-balanced is good enough, right?) and don't want to pay for rockers or custom cam lobes that match the motion ratios of their valvetrain.

It's all a game of trade-offs. Define what the car needs to do, and the PLAN becomes fairly obvious. It's the 'budget' aspect that starts to make the 'streetable' argument a constant validation for reigning back from what's intended.