I would like to report that I ended up with a better than 3-2-1 solution that was very different from what I sought out to do in the fist pass. If you are interested in how I did it, please read on.
First off, the attempts to upload Time Machine backups to a cloud data store was a bit of a catastrophe and frankly a bad idea from the start. I just couldn't efficiently upload the sparse bundles for some reason. Plus, I realized that with a 4 TB TM drive performing a restore would have been a nightmare. I was using Arq as my backup client which is normally blazingly fast with its multi-threading and over-the-wire compression. For whatever reason, it took Arq a very long time to scan the Time Machine backup data so upload speeds were always abysmal. I chalk this up to the nature of the TM data store and not Arq, which I can say after a thorough testing is otherwise rock solid, secure, extensively featured, has flexible overhead, and at top speed will probably smoke any other all-in-one provider. The big plus with Arq is that it supports so many cloud data stores so that regardless of the size of your data set, speed requirements, and data integrity needs, you can pick the right provider for you rather than get locked into a single vendor. Plus, it also supports SFTP so you can backup data locally or to a non-native cloud store like Backblaze B2 if you really wanted to.
Because I ended up choosing to have duplicated Time Machine backup drives (which I will swap out every week so one is offsite almost all of the time) and because I have several TB of data, I ended up choosing Amazon Cloud which is about $60/year for unlimited files. With my 30 Mb uplink I can saturate it and then some thanks to Arq's compression. I would probably only need to restore from it if there was an earthquake that knocked out both of my Time Machine backups, or to "top" off files if I had to backup from the offsited TM drive. Also, if there are concerns about data integrity over time, Arq has the feature to vigilantly validate the backups to make sure there is no data rot. If something fails a checksum it's dumped and uploaded again. If it is not available, you are told exactly what was lost. I now have my bases covered. I don't really need backups to the beginning of time but, saving corruption (I don't know how well things "keep" on Amazon Cloud) I will have what I need, which is up to a month's worth of history.
So, in the end I chose backing up my system directly rather than attempting the backup of a backup, which I realize now would have been fraught with peril anyway. Arq's clever techniques of avoiding the backup of unnecessary files seems to put it on par with what Backblaze does. When you restore, you still have to reinstall your apps, but all of your Library data (minus temp files) will be intact. I am backing up 2 hosts separately along with 2 TB of media. It is tearing through the first backup right now at around 3-4 MB/s and at this rate I'll be done by early next week. Cost-wise (and since I already owned Arq) this makes the most sense for me.
In conclusion, this approach ended up being the most convenient and cost effective way to provide 3-2-1 level backups of my data. Also, with much of my data on iCloud or DropBox it's actually more than that. I love the convenience of Time Machine, especially when it comes to restores. It's a one and done solution that will fully restore my system to what it was if I have to replace my Mac. The swapping of drives is a bit manual, but it's only once a week and I have a nice drive case for transport. If I have a fire or theft and am forced to restore from the slightly older offsite drive, I know I have the cloud backup to get my system up to current where necessary. Still, the cloud is mainly for peace of mind in case things really go sideways. I hopefully will never need it, but if I ever do, I will be glad I spent the small amount for it. I hope that documenting this entire exercise will be helpful to others.