I was also wondering about this. There should be some sort of hierarchy on how files are processed through the scanning/protection pipeline. If you have well defined hierarchy, you can't get such conflicts because when protection module in the beginning of the protection pipeline confirms a detection, continues process is stopped and file is quarantined.
They should place DeepScreen in front of Software Analyzer and also include Software Analyzer within the virtualized DeepScreen. So it would initially inspect the malware using DeepScreen AND Software Analyzer WITHIN virtualized environment and then release it to host where it would be again monitored more extensively with Software Analyzer on an actual live host system. I mean, I've seen ransomware during testing (not necessarely during testing of avast!/AVG) which was detected, but after it has already encrypted files. If it tries that inside virtualized environment, you'll still catch it without giving it a chance to do any harm.
Running Software Analyzer on host level as well is a good idea because malware can refuse to do malicious actions when it knows it's running in a virtualized environment. Or within a short timeframe of DeepSreen monitoring.