I don't have any answers to your questions because no products exist yet that can do this. But I'd like to comment anyway because around 10 years from now your post will mark a historical point in IT infra monitoring/management history.
This set of problems you have defined is exactly how the cloud paradigm shift is being led away from large heavy servers and large, heavy monitoring applications.
This is why you do not see any advanced, robust monitoring platforms for AWS yet that are as in depth as SolarWinds. Nagios and DataDog are not it.
Heavy monitoring is not important enough for rapidly built and destroyed servers.
In the old world many of us old IT dinosaurs live in now, we would call Microsoft when chits broken. In the AWS world of the future, servers are terminated and rebuilt with auto scaling without any thought to troubleshooting. All of us dinosaurs have servers that are like old friends, the thought of wiping them and re-creating them in real time is a horrific thought. This is the heaviness of the infra world we are used to. The 5-10 year old 2008R2 servers in the racks or VMs that still power Susie's old accounting apps are dinosaurs just like us. AWS wipes all of this away.
The new world is not heavy and will not be managed by big heavy apps. A new kind of Elastic Monitoring platform will need to be built on AWS that responds to these changes in infrastructure dynamically, using all of the combined :::arn::: AWS resources. I suspect SolarWinds has a team of people building an AWS centric tool using Lambda and all of the other goodies now as AWS will be taking over the entire world in the next 10 years and make Microsoft, Cisco and VMware nearly obsolete.
Just my ,02 cents because I also need everything you're asking for, and if someone made an AWS marketplace AMI to do all that SolarWinds does but one that is Elastic, and follows all of the AWS principles of DevOps, and dynamically monitored all in the customer virtual private cloud, it would be an epic money making machine.