timeout: `0x1d` should be parsed as 29 seconds, not 1 day

[drinkcat]

Inspired by this discussion: https://lists.gnu.org/archive/html/bug-coreutils/2025-04/msg00008.html

$ strace -e trace=clock_nanosleep sleep 0xad
clock_nanosleep(CLOCK_REALTIME, 0, {tv_sec=173, tv_nsec=0},

Here 0xad is interpreted as 173 seconds, not 10 days.

Given FreeBSD and ksh sleep implementations also behave like GNU,
we should probably just document the edge case that
(day) suffixes are best avoided with hex inputs.

We currently parse 0x1d as 1 day, not 29 seconds.

[drinkcat]

Also fixed by #7675, but we don't really catch this exact case in a test AFAICT. We could do end-to-end tests on sleep or timeout with a value like 0x0.1d (that should be ~0.11 seconds, not 1/16 of a day = 1.5 hours).

@Qelxiros FYI as well. Again, I'm happy to look into this, either way is fine.

[Qelxiros]

Same as the other one. I'll look in a day or two if you haven't. Just commenting so I don't lose this issue. Thanks!