(In this context, self-hosting has two parts: the ilo computer
and the ilo.rom)

Due to running on a virtual computer, Konilo is not currently
capable of being self hosted. But that could change. As a first
step towards this, the assembly code for the ilo.rom is now in
my blocks, and I've started exploring adapting the embedded
assembler to be capable of building a new image from this.

The tricky part here is updating the actual ilo.rom. Saving this
is (on ilo systems supporting it) just a matter of `rom:save`,
but this just writes the memory to the image. So the newly
assembled image needs to be relocated over the old one before
this is done.

It's something I've done before: RetroForth 11 was built this
way, and it worked ok, at least in the early days. (Creeping
complexity ultimately proved to be a problem there; something I
am trying to avoid repeating with Konilo).

So I'd say the first part of this is doable. It won't be a quick
process, as, due to health concerns, I need to avoid pushing too
hard, but it is something I can achieve.

The second part, replacing an ilo computer, is harder. A couple
of years back I didn't see a good path forward on this, but jmf
wrote an x86 assembler and then a nga implementation in it. This
clearly showed that it is possible, and with less difficulty
than I had expected.

I think that, for older x86 systems, it'd be possible to go from
our current state to a self-hosted state within a couple of
months. (With some additional help it could be much faster, but
again, my ability to type for more than a few minutes at a time
is not sustainable anymore).

In 2024 I'll begin work on a port of jmf's RASC to Konilo as a
step towards the self-hosted ilo. Getting this fully working
will certainly require additional contemplation as new issues
arise, but it'll make for some interesting challenges as work
towards this slowly progresses.

And I do mean slowly: I'm ok with using an ilo + ilo.rom built
separately.