fixes #20015; document shallowCopy does a deep copy with ARC/ORC

[ringabout]

fixes #20015

[github-actions]

Thanks for your hard work on this PR!
The lines below are statistics of the Nim compiler built from 10c8e20

Hint: mm: orc; threads: on; opt: speed; options: -d:release
163320 lines; 14.970s; 838.762MiB peakmem

[arnetheduck]

@Araq is this the long-term plan, that shallowCopy is broken from ARC/ORC? ie this is a significant semantic change - if it is the case, it should really be deprecated / removed-when-arc instead - these bugs will be incredibly hard to find

[Araq]

Yes, you're right. We should simply remove it from system.nim for --mm:orc.

(You're supposed to use x = move y instead if the compiler does not know a move is valid.)

[arnetheduck]

(You're supposed to use x = move y instead if the compiler does not know a move is valid.)

this is semantically different: a shallow copy allows (mutable) aliasing (in an unhealthy way, I'd argue, but this is a different story) - so the two are not .. equivalent - that said, good riddance.

[Araq]

I know it's semantically different, else we would implement shallowCopy as move... But often it can be used instead.