It is kinda disappointing but I can see why Qualcomm wants to use the brand.
More specifically, I can see it trying to compete with things like those funny Chinese boards built around SoCs like SG2000. Those embed a Linux capable core, a small NPU, a camera interface with ISP and video codecs, and a secondary RTOS core for realtime control. Basically built for drones and simple robots. The caveat of those boards being: the documentations sucks, the SDK is wack, you get 3 example scripts and are entirely on your own outside that.
Qualcomm could be trying to branch into drones/robotics/etc with this move.
The concern I have with the $44 Q is it has 2GB of RAM and 16GB eMMC, and a processor that's probably between a Pi 3 and Pi 4 in terms of speed and IO (though 4nm, so probably much more efficient).
For $45 I can buy a Pi 5 with it's own built-in GPIO, PCIe, and a much faster SoC, though it lacks a few niceties like the Q form factor, the more efficient SoC, a realtime microcontroller, and a USB-C port with display out capabilities (I really wish Pi had that...).
But what you think of as an old core design is in fact a mature, well-understood, well-tested, widely-supported, cost-effective core design. It also has some features such as in-order execution which none of the newer chips have. From an engineering perspective, it still can make a lot of sense in the right applications today.
This _is_ possible with Linux, but not at all trivial and likely impossible with general-purpose distros.
Interrupt handling and (on RP2040) dedicated multicore code is also nice.
I assume initramfs-only with special purpose pid0 and only the modules needed statically compiled into the kernel?
What else would it take?
Bootloaders need to initialize most of the devices and load the kernel image. Then they hand the control over to Linux which proceeds to re-init these devices again.
The userspace matters, but on recent computers it doesn't matter that much. You can get to sub-40ms with https://katacontainers.io/ That's a project that uses full VMs to run Docker images boot instead of kernel namespacing.