fixes |mass

[joemfb]

refactors and unifies memory marking implementation for GC and |mass

[philipcmonk]

Just a couple of comment suggestions

[joemfb]

I'm testing these printfs further, making sure they work with arbitrary structures and large amounts of data. I'll push some fixes soon.

[ixv]

Can we save this output to a file? It would work nicer with this: urbit/arvo#1028

Maybe we should #define LOGGING or something for things like this.

[belisarius222]

I feel like ideally we'd be uploading this data to https://docs.datadoghq.com/graphing or something similar.

[belisarius222]

Did a somewhat cursory read of this. Looks good.

[joemfb]

I cleaned up the comments, and disabled printing of atoms larger than 10 words. The allocation-box heuristic printing is still crummy, but it's quite a bit better than it was.

@ixv, I could thread a File* pointer instead of a loobean through all the mark functions -- that could default to stderr but pointed to some external file through a build option or something.

[joemfb]

@ixv, configure meson with -Dmemory-log=true and every |mass effect will be written to pier/.urb/put/$DATE-mass.txt.

[joemfb]

@philipcmonk, please re-review.

Resolved.