[llvm] [yaml2obj] Apply output size limit to Mach-O (PR #210027)

[via llvm-commits]

AI-Risk-Management wrote:

Thanks for asking. The motivation was an internal review of consistency in `yaml2obj`'s output-limit handling after #209695: ELF and COFF enforced `MaxSize`, while Mach-O did not, and Mach-O's `ZeroPadBytes` path constructed attacker-sized temporary vectors.

I used OpenAI Codex substantially to identify, implement, and test the change; the `Assisted-by` disclosure was intended to state that. I reviewed the patch and test results, but the implementation and PR description were substantially tool-generated. Having now read the policy more carefully, I understand that this is not the contribution model LLVM wants and I should not have opened it in this form.

Please keep it closed. I will not ask for further review or reopening. Sorry for the review noise.

https://github.com/llvm/llvm-project/pull/210027