A project I mentioned earlier was to (again) refactor the exec-path-from-shell Emacs library. In thinking through the approach, I've come to the perspective (where I think I've landed before) that the library should be unnecessary and should instead the issue should be resolved upstream. While its use offers a pragmatic alternative, in my particular case I'll look to adjust the invocation as necessary on systems where it presents an issue.
To some extent this raises larger questions around the relationship between shells and processes which may be invoked more directly which I'll also pick at a bit, and may be somewhat addressed through adoption of tools like Nix.
I had somewhat assumed that my server would support UEFI and so optimistically followed those instructions. After being unable to boot I did some quick poking around and discovered that the server is an HP Compaq 6000 which is a few models short of when UEFI support was added and there's seemingly no firmware upgrade(1). This is a bit surprising but not overly so since I picked up the system refurbished several years ago where I think my primary criteria was relatively standard components and decent storage for media. Before having settled into a house I'd mostly transitioned to just using laptops for their smaller footprint, and as much as anything I think this server was intended to serve as a starting point where I could replace components as necessary (where the whole system was a comparable price to getting a decent case and power supply alone).
After swapping over to the legacy boot option, NixOS is running and now just begins the process of configuring it which I'll probably initially do primarily over SSH. As part of this I'll probably also audit my home network a bit.
After creating a standard user in NixOS, I'm able to ssh into the server so I can start to capture the configuration externally as configuration.nix.