Given that airflow is handled under fluid dynamics, wouldn't the amount of flow and direction changes matter far more than blockages? It's like how Linus Tech Tips threw a ton of crap into a case to see how much it took to make an appreciable difference on temperatures based off the old wives tale that bad cable management would affect temperatures. It's also the opposite problem that I had with my old Thermaltake Core X9 case. That case is big and is meant for water cooling, and that's how I used it. Although, I still had fans to help provide some flow to motherboard components and such, but the M.2 drive had massive temperature spikes (easily hitting its limit of 90% of max TDP of 70C). The reason? The case was too voluminous for the amount of airflow/pressure. At one point, I just stood a 120mm fan upright near the bottom of the motherboard, and that helped tremendously.
the blockages in this case is affecting total volume of coolant.
so cooling in the case of air coolers is a function of
1)heat output as a function of surface area of the sink and the transfer rate of the material its made out of.
2) the amount of coolant fluid and the temperature difference between it and the sink.
as a hypothetical to demonstrate:
you have a soc on a raspberry pi sized pcb to which you attach a tower style cooler with a 9cm wide x 9cm deep x 9cm tall heatpipe finstack made of aluminum. you then have a 10cm x 10cm x10cm cube cup filled with 20 degC water. you flip the pcb upside down on a rig that can lower just the finstack into the cup of water. you run the soc and lower it into the water for some time increment. the finstack transfers the heat output to the water raising the temp to 50c. you raise the fin/soc/pcb up and a 2nd cup of 10x10x10 of 20c water comes down the conveyor belt and you repeat the water heating process.
the 10x10x10 cup of water is 1000 cubic cm or liter. that liter of water at 20c has a fixed number of kCalories it can absorb from the sink finstack. if you reduce the height of the cup and sink by 2/3, the cup is (10x10x3.3) and the finstack is (9x9x3). you have reduced the amount of coolant and thus the thermal capacity of the coolant fluid by 66% and the thermal transfer rate also goes down since you have less fins and less surface area. you could change the finstack material to copper and gain 40% more transfer to make up for the surface area loss but it is more expensive and heavier.
in this example the cup represents the interactible volume the cooler has with the coolant fluid. reducing the size of the volume has a directly proportional affect on the transfer rate and the upper limit of the coolant to absorb heat.
[it is the difference between throwing a 12oz can vs a 1liter bottle of soda on a hot skillet. which one is going to cool it more?]
by going from the 120mm x120mm cross section of the se:X to the maybe 100mm x 25mm narrow passageway, ms has reduced the max interactible amount of fluid.
lets go back to our inverted sink setup. we've reduced the height of the sink and cup to 1/3, but to make up for the cross section loss we can extend the length to 3x to compensate. so we make the heatpipe array in to something like the noctua nh d14 with pipes running to 3 separate finstacks with 9x9x3 size each (total dimensions 9x28x3). we set up our conveyor belt with 10x10x3.3 cups spaced so the 3 finstacks can fit into 3 of them at the same time.
we turn on the soc and we start transferring heat to cups 1,2,3 heating the water from 20c to 50c in each. we lift the rig, advance the conveyor line, lower the rig, and start transferring heat. cups 2 and 3 go from 50c ->80c while cup4 goes from 20->50. we advance the line again, lower, and run the soc. cup 3 now goes from 80c->100c and hits the thermal absorption limit of the water, cup 4 goes 50->80, cup 5 goes 20->50.
this is the problem with long path heatsink setups(blower gpu cards). some parts of the sink gets to work with 20c coolant while other parts only interact with 80c coolant. you have to run integral math to get the finer gradations but the principle holds.
by lowering the internal volume of the air passageway of the se:X the coolant becomes less effective. now water as a fluid is uncompressible and stickier than air so the dynamics are different. and that is a good thing, because it allows you increase coolant flow rate to make up for the coolants reduced ability to absorb heat. but ramping the fan rpm is probably going to be limited to keep noise down.
so assuming you are committed to the form factor, you have other options. increase the depth of the single 120mm fan to increase airflow, stick another 120mm push fan on the other end to increase airflow, or add ducting and extra length to take the 120x120 passage and squeeze it down along a gentle curvature with laminar flow to take advantage of bernoulli's principle to increase airspeed at the finstack and an expansion section to the ducting at the exit to take advantage of the heat expanding the air volume in a quasi scavenging effect. MS did none of these things.
instead of double stacking the motherboard and the io board on a 30mm thick chassis, they could have mounted the board edges at 90deg forming a L that lined 2 sides of the case and opened up more of the internal volume for more air/coolant.
the air path matters but it is the air temp and speed at the sink that will determine how much heat is rejected from the system. at idle the se:X in a vertical orientation is a great design for a passive cooler situation taking advantage of the stacking(aka chimney effect)cooling with the fan forcing the effect in the horizontal orientation. but again this is not a great setup for dusty environments.
as for your thermaltake x9, yeah running all water would be an issue. buildzoid once went over the watts generated by the other components around a vrm dimm and m.2 slots as an aside. anything with thru hole pins was passing 3 to 10-ish watts (or some surprising number)to the copper layers in the motherboard. that can add up with all the smd components around a module.
instead of 1 big 230mm fan at the front of the case what they needed to do was mount 4x 140mm fans in a 2 by 2 square on the front and back of the case to make a single flow that filled the entire cross-section. there is a thing in cooling where you go for total volume replacement every couple of seconds. so every cubic cm is pulled out with no deadspots or nooks where eddys could form. this requires very minimal blockage. i guess you could have the 2by2 140mm chamber for the motherboard and the rads in a separate chamber with their own fans. i've seen some custom cases with front to back flow.