Fix IO graph

[kha84]

Hello there!

I wanted to revive this one - #41 (comment)
I do have almost everything together to get it working, just one minor thing has left - to properly collect process-level IO metrics, as now it's different from what I see in other IO monitoring tools (like iostat).

Screencast.from.2026-05-04.12-45-56.webm

What was fixed so far:

• asciigraph crash on all-equal data
• device name mismatch between /proc/mounts and /proc/diskstats
• incorrect scanf field mapping for /proc/diskstats
• IO graph displaying raw unscaled bytes

Do you mind me raising a PR when it's ready?

[kha84]

OKay, there're no issues with collecting per-process IO statistics, apart from the known ones - which are permissions. When the ttop is running under user, the procfs doesn't let it to get /proc/[pid]/io for processes user doesn't own. And solution for that is obvious - to run it under root :)
I'm happy to contribute my patch, if you're willing to review it