I don't think you'll miss anything. If pihole works for you, then there is no need to switch to adguard.
One thing I found helpful is configuring my router (asuswrt-merlin) to transparently route all dns request to my adguard instance. You might already heard that some apps and IoT devices tried to be clever and hard-coded their dns server so they can evade dns blocking (I'm looking at you Netflix). If your router support redirecting all dns request to a custom dns server, definitely use it!
Sounds great! By the way, if you're using docker, be careful not to accidentally have a container open a port on all interface. Even if you have a firewall configured on the machine, sometimes docker can punch a hole without you knowing. Might be a good idea to run a port scan from an external computer from time to time just to makes sure no unwanted open ports.
Unfortunately, I can't say for sure if your instance will reliably get the deletion requests and process them. I did a small test to see how deletion works a few days ago and it doesn't seem to propagate reliably as the deleted comment is still up in another instances, even now, though other instance such as lemmy.world seem to delete it. Not sure where it went wrong either, could either a bug, instances get overloaded and didn't receive activitypub message correctly, or OP's instance was improperly configured, but I sure hope it's just an isolated incident.
Like I said before, If this still worries you, you can just delete older image files in the pictrs directory every few months to make sure you don't host user-uploaded files for too long.
If you look at the chart, pretty much nothing comparable to the same period last year. January 2023 is a lot higher than January 2022 for example. July 2023 is also higher than July 2022.
Some people say manually purging the activity table for entries older than a week or so should be safe enough.
The risk is pretty small IMO, especially if you (or your friends) are the only one that use your instance (with registration closed so no random users uploading stuff to your own instance). If you disable nsfw on your instance, the chance of storing illegal images should be pretty low, especially if the communities you subscribed are moderated as deletion from mods will eventually processed by your own instance. If this still worries you, just nuke pictrs directory every few months, perhaps automatically using a cron scripts that delete images/gifs older than a few months.
If you compare post per days from before the strike, it definitely falls. It's no longer an upward trajectory despite subscribers growth.
If you look at the charts you linked, you can see the users activity (post per day and comments per day) is falling sharply since last month. Subscribers count mean nothing if a big proportion of the active posters leave.
I'm self-hosting an instance of voyager and it's been great so far. The devs are pushing updates so often, sometimes I have to update it twice in a single days.
This seems to be the easiest solution. Use the provided docker-compose file, then configure traefik to route requests to your lemmy domain to port 8536. How to do that depends on how you currently run traefik as there are multiple ways to configure it. Could be as simple as adding a label to the service named proxy in lemmy's docker-compose file.
Microsoft is probably considering to release an enterprise Linux product right now. Perhaps called Windows Subsystem for Enterprise Linux.
Nvidia cards are mostly working fine these days as long as you're not using Wayland. If you're using Wayland, be prepared to encounter lots of minor annoyances, and perhaps some bugs that completely break your workflow depending what you're using Linux for (e g. on server you don't have to deal with sleep issues, but in desktop it's an annoyance while on laptop it might be a deal breaker).