A whitelisting application has a list of what it knows it bad AND what it knows in advance to be good.
How would it know this? Is this defined by a person/people? If so, that wouldn’t have mattered. liblzma was known in advance to be good, then the malicious update was added, and people still presumed that it was good.
This wasn’t a case of some random package/program wreaking havoc. It was trusted malicious code.
Also, you’re asking for an antivirus that uploads and uses a sandbox to analyze ALL packages. Good luck with that. (AVs would probably have a hard time detecting malicious build actions, anyways).
It places unknown/new software in a sandbox. You want an AV that tests all pre-existing packages in a sandbox.