I’m selfhosting a Matrix server and have all my Chats from other apps also bridged to there.
Same here, but with XMPP in place of Matrix. For historical context, XMPP was invented about 25 years ago on the premise that people were already tired of having their instant messaging scattered over multiple protocols (rather than Signal, Telegram, WhatsApp, Discord, iMessage now, it was Yahoo, MSN, AIM, ICQ, … then), so bridging is very much front and center in the XMPP world. Over time, people also realized that bridging sucks in general (you either dumb down your client to the lowest common denominator which sucks for yourself, or your client isolates itself from the source protocol enough that it sucks for everyone else).
To add insult to injury, most modern protocols also forbid, by their ToS, the use of alternative clients (which very much includes bridges), and to the best of my knowledge WhatsApp, Signal and Discord will eventually suspend your account on this basis.
Matrix is still trying to carve a niche for itself in this space, and is failing IMO (judging by the quality/security of the bridges they have come-up with, and the recent libera.chat fiasco). I'd say that the situation in this regard in XMPP is only marginally better due to the fact that XMPP had a decade headstart to fail and try over, and I would not recommend using bridges on either of them if that can be avoided.
It XMPP better for group VC?
I'd say "it depends". Fun fact, Matrix uses jitsi-meet under the hood (which is XMPP + a media transcoding/multicasting component that doubles as a relay), and jitsi-meet is my recommendation for this use-case: as long as the central server has good bandwidth, you can really scale up your VC to many attendees. On top of that, XMPP has support for peer-to-peer group VC, with the benefit that hosting is simpler, it doesn't require any central component/relay (but the bandwidth cost is incurred on all participants and you won't go beyond a handful of attendees that way).
It's bad omen to criticize an opensource project, I know, but in my eyes Matrix is a big technical and organizational failure, for not having succeeded in stabilizing the protocol after a whole decade of unsuccessful explorations, and for having its leadership consistently fail to define clear goals and steer the project towards them (just get it done and working well before trying to make it "peer to peer" or "in the metaverse").
If this is the electroshock that Matrix needs to reconsider its design and directions? good for them. If that kills them? Well too bad, but it's not like they are the only cool kid in town.