Interesting! That's very close to this blog post I read long time ago (unfortunately medium.com link)! Are you actually sending emails from those addresses? Like if you need to drop an email to your bank, do you use the banking one or your personal (or something else)?
Fwiw, I do something similar. I use a mix of domain aliases without address (e.g. [email protected]) and actual aliases. Since I have proton family (and the same when I used ultimate) I have unlimited hide-my-email aliases, so I have it integrated with my password manager, and I generate a random password and email for everything I sign up now. These though are receive-only addresses. In fact, with this technique I probably use 3-4 addresses in total, but I have probably 30 domain addresses that go to the catch-all one.
Spam on these addresses are basically non-existing and you can still create folders based on recipient without having a full address (e.g. [email protected], [email protected]). You can make folder categorization based on recipient regex and this way you also have the "stop bothering me" option: if some email gets into the wrong hands, you can create a spam rule for that dedicated address. However, my approach is that all of these are used just to receive emails, to send I have just a handful of actual addresses or -if really needed- I can create on-the-fly an address from a catch-all one, send the email and then disable it again (so it doesn't count towards the limit, but I still get inbound email to the catch-all).
Nice setup anyway!
You should definitely be! I take backups every 6h for my self hosted vaultwarden (easier to manage and to backup, but not official, YMMV). You can also restore each backup automatically and have a "second service" you can run elsewhere (a standby basically), which will also ensure the backup works fine.
I have been running bit/vaultwarden now for I think 6 years, for my whole family and I have never needed to do anything, despite having had a few hiccups with the server.
Don't take my word for it, but the clients (browser plugin, desktop app, mobile app) are designed to keep data locally I think. So the term cache might be misleading here because it suggests some temporary storage used just to save web requests, with a relatively quick expiration. In this case I think the plugin etc. can work potentially indefinitely without server - something to double-check, but I believe it's the design.