fix: raise AVM check-circuit resources

[AztecBot]

Summary

• Increase the AVM dumped-input check-circuit command resource prefix from the default 2 CPUs / 30s to 4 CPUs / 60s.
• Keep 16-way parallelism within the 64-core CI host capacity while allowing larger valid AVM traces to finish.

Root Cause

The failed AVM check-circuit run did not hit a relation failure. The isolated multiple_blobs input was killed by the harness timeout: it reached Running check (with skippable) circuit over 700560 rows and then timeout sent SIGTERM at the hard-coded 30s per-input limit.

Testing

• Reproduced the pre-fix command generation for the failing input path and confirmed it emitted ISOLATE=1:TIMEOUT=30s.
• Re-ran the same generator after the change and confirmed it emits ISOLATE=1:CPUS=4:MEM=8g:TIMEOUT=60s.
• Ran git diff --check.

I could not locally rerun bb-avm avm_check_circuit for the failing input because this workspace does not have the cached input artifact or a built bb-avm binary.

---

Created by claudebox · group: slackbot