So you want 10-35y of experience in a few sentences ..
> Why is it impossible to tweak audio drivers that they get priority over network,
> video and storage subsystems [in windows]?
Windows is not a real time Operating system. What I understood so far. When a device driver receives an interrupt it starts a DPC (deferred procedure call). This runs on a core and can't be interrupted by anything else (process scheduler, etc). Its only a convention, that device programmer know, that these non interruptable driver routines shall run not for too long, too avoid blocking the kernel for too long for other also important tasks. If there is a bad driver which doesnt give the core free early enough and your audio driver and/or DAW application runs on the same core, which is blocked by this driver for too long, then you have a typical situation why dropouts can occur or why you require a bigger ASIO buffer size to mitigate this.
Other reasons of bad audio performance are, when you use energy saving settings in the BIOS and in the Operating System. They lead to 2 things. Either the CPU frequency changes depending on load or parts of the CPU are being send to sleep state. Both, frequency changes and sleep states (C-States) lead to delay as it takes time to synchronize on a different clock speed. And waking up parts of cores takes even much more delay. So its a good idea to at least set C-States to perhaps C0, disable EIST, and then see how far you come. If all runs satisfactory you can try to re-enable EIST, Turbo boost and decide whether you want to allow maybe up to C3 sleep state.
Whats also not beneficial is the feature CPU core parking as this also leads to a more laffy behaviour of the CPU. It takes time, until the parked cores are back there.
Other reasons for bad performance can be the mainboard layout and the Chipset in use. Nowadays the PCI slots at the end of the day also terminate on an PCIe Bridge ... so data from a parallel bus design land on a serial bus design (PCIe). Depending the PCIe Lane distribution on the mainboards this can be a good or a bad design.
Some technical background you can read on the homepage of LatencyMon. Read this information 3-5 times. Until you maybe start to understand what the issue is.
http://www.resplendence.com/latencymon
Some performance tuning tips I gave here already in a different threat.
https://www.forum.rme-audio.de/viewtopic.php?id=22948
On a perfect tuned Windows 7 system you will see with LatencyMon on an IDLE (!) system, where no other program runs, values down to 1,75µs. Usually between 5-10µs with little spikes up to perhaps 50µs.
This gives you the Headroom, being able to work longer with small ASIO buffer sizes, shall you require them for some reason, i.e. when playing an electronic drumkit and if you have a virtual instrument on the PC.
Or if you use playback functions of the DAW instread of doing the headphone mixes purely on the recording interface.
BR Ramses - UFX III, 12Mic, XTC, ADI-2 Pro FS R BE, RayDAT, X10SRi-F, E5-1680v4, Win10Pro22H2, Cub14
