Unsafe execution of untrusted code via `exec` by default

[jamesmurdza]

The library directly calls exec() directly on model-generated code. The warning notes this should be commented out by default, but it isn't.

human-eval/human_eval/execution.py

Lines 40 to 50 in 6d43fb9

	# WARNING
	# This program exists to execute untrusted model-generated code. Although
	# it is highly unlikely that model-generated code will do something overtly
	# malicious in response to this test suite, model-generated code may act
	# destructively due to a lack of model capability or alignment.
	# Users are strongly encouraged to sandbox this evaluation suite so that it
	# does not perform destructive actions on their host or network. For more
	# information on how OpenAI sandboxes its code, see the accompanying paper.
	# Once you have read this disclaimer and taken appropriate precautions,
	# uncomment the following line and proceed at your own risk:
	exec(check_program, exec_globals)

The best solution would be to provide at least one sandbox integration as a reasonable default. I'm open to contributing this.

@mpokrass