Run c_crtl_wait.c on borg under qemu-riscv32 10.2.1

The whole task is one compile and one run, and its value is entirely in where it runs.

./compiler/pascal26 --target=riscv32 test/c_crtl_wait.c -o /tmp/<scratch>/out
<path-to>/qemu-riscv32-10.2.1 /tmp/<scratch>/out

diffed against borg's own gcc oracle for the same source, looking at the wait4-rusage line specifically (rusage=written versus rusage=UNTOUCHED).

Why this is not "just re-run the failing row"

The row already fails on borg under its installed qemu 8.2.2, and has ~700 times. Re-running it produces row 701 of a table that cannot answer the question, because host, gcc version and qemu version are perfectly collinear in every observation this fleet has ever recorded. The empty cells are the informative ones and no amount of ordinary sampling fills them.

This task fills one: same host, same kernel, same gcc, same pxx binary, emulator moved by hand.

Constraints, and they are the reason this is specified rather than assumed

• Invoke a 10.2.1 binary by path. Do not install, upgrade, or change a package. A dist-upgrade is sudo, therefore the owner's, and it is also not what the question needs — the question needs one process run under one emulator.
• Do not satisfy this on plexus or any other host. frankb-8e already ran that arm: GREEN, qemu=10.2.1, gcc=15.2.0, tree 3e381e41d. It lands in the cell seven already occupied, so it REPLICATES that cell on a second machine rather than adding one. It does eliminate "something about seven the machine", which is worth having and is recorded in the A ticket; it does not touch gcc.
• Report the emulator's identity, not the intent. Read the version off the binary actually executed. tools/run_target.sh riscv32 is a bare exec qemu-riscv32, i.e. a PATH lookup, so which binary runs is an environment fact and not a property of the command.

What to do with the answer

Write the verdict into bug-a-wait4-does-not-write-rusage-on-riscv32 (backlog-core, p55) and say which qemu binary produced it. That ticket already carries the full analysis and both outcomes are written up there; this one exists only because the measurement is unreachable from where every seat currently sits.

Do not re-rank the A ticket on the strength of a green here without reading its last two sections — its prio is justified under both outcomes, for different reasons, and that is stated there deliberately.

Provenance

Collinearity counted by frankz-e5 over devdocs/progress/tstate/reports/*.md; plexus arm measured by frankb-8e; the emulator-split hypothesis and the seven-retirement timing are frankb-8e's. Neither of us has run this row on borg and neither of us can.