rootfs: iSH-AOK is no longer a host that rejects the chroot mounts

[emkey1]

Second of two PRs from the iSH-AOK side. Independent of #1, either order is fine.

What is out of date

rootfs_mount_chroot_fs carries this note:

Restricted hosts such as iSH-AOK may reject one or more mount types

That was true when it was written. It is not true now. As of iSH-AOK build 548 all build types are supported, and I verified all five mounts this function makes on an aarch64 iSH-AOK guest:

mount -t proc    -> OK        mount --bind /dev      -> OK
mount -t sysfs   -> OK        mount --bind /dev/pts  -> OK
                              mount --bind /AOK      -> OK

with /proc populated and /AOK showing real content afterwards. On our side that is bind mounts and MS_MOVE landing, devtmpfs becoming a real filesystem rather than a no-op, and proc/sysfs/devpts/tmpfs/devtmpfs all being registered mountable filesystems.

What this PR does not do

It does not remove the fallback. Older iSH-AOK builds are in the field and other restricted hosts exist, so the try-and-degrade structure stays exactly as it is. The only thing removed is the claim that failing is the expected outcome on iSH-AOK.

What it adds instead

The warnings become actionable. On an iSH-AOK older than 548, a failed mount now says so and names the build that fixes it:

Could not mount proc in /tmp/x — iSH-AOK build 547 predates full mount
support (build 548+); the build continues without it.

Anywhere else the message is unchanged, since the generic wording is all that can honestly be said.

Build detection reads the number out of uname -v (iSH-AOK 1.3 (547) built ...) rather than the marketing version, which we do not bump between builds and which therefore cannot tell two builds apart. That is a trap worth knowing about generally if you ever key behaviour off an iSH-AOK version.

Testing

Exercised by sourcing the module with tui_* stubbed, on four hosts/cases:

case	result
real iSH-AOK guest, reports 547	names build 547 and the 548+ requirement
simulated build 501	same
simulated build 548	generic message, no misleading upgrade advice
malformed uname -v	generic message, no error
non-iSH host (Darwin)	generic message

bash -n clean. tests/*.sh give identical results before and after (test-audit-fixes 3 failing, test-features 49 failing, other three green; both of those also fail on a pristine 5a0b3fd, so they are unrelated to this change).

🤖 Generated with Claude Code