I think the supposed risk to electronic voting machines is that there would need to be thousands of them, are distributed, somewhat unattended, and operated by people that don't know them.
The possibility of an exploit or misconfiguration increases, and the ability to compromise someone supervising one of the polling station increases.
If there is are centralised systems, fewer higher skilled people would be required to secure/monitor/run the system. It can also be airgapped.

While some of these risks are also applicable to in-person and mail-in voting, these systems have been around for ages, are not proprietary, and anyone can figure out "how it works" and can make sure "how it happened" matches.
As soon as you get into cryptographic vulnerabilities and security, 99.99% of people would be lost in the woods

The rest of the questions, I feel, are more systematic things.