[bxyu-nvidia]

What does this PR do?

1. Enable async scheduling again: benchmarks/nemotron_3.5_super/vllm_configs/nemotron_3.5_super.sh
2. Swap to PTY: resources_servers/swebench/app.py, resources_servers/terminal_bench_2_1/app.py, responses_api_agents/opencode_sandboxed_agent/tests/test_app.py, responses_api_agents/opencode_sandboxed_agent/app.py
3. Fix Terminal 2.1 golden patch: resources_servers/terminal_bench_2_1/app.py
4. Misc cleanup: benchmarks/nemotron_3.5_super/sbatch_external_vllm.sh, nemo_gym/rollout_collection.py, nemo_gym/server_utils.py, resources_servers/swebench/configs/swebench.yaml, resources_servers/swebench/swebench_patches.py, resources_servers/terminal_bench_2_1/configs/terminal_bench_2_1.yaml, responses_api_agents/opencode_sandboxed_agent/configs/opencode_sandboxed_agent.yaml

Checklist

• I have read the contributing guidelines.
• The change is focused; unrelated "drive-by" edits are tracked as separate issues/PRs.
• Tests added or updated and pass locally, or N/A for docs-only / non-code changes (so CI unit/server checks pass when applicable).
• Pre-commit checks pass locally (pre-commit run --all-files) (so CI lint/format/copyright pass).
• All commits have DCO sign-off (git commit -s) (so the DCO check passes).