the question about why not put the whole frontend into the extension is an interesting one. basically, this is a tradeoff about security vs decentralization putting the FE in the extension means enshrinement, but it's important that there be a wide variety of third party frontends, so that no single party is in control of the user experience. if the first FE was put in the extension, this would create an assumption that that's the way things should be done, and other FEs would be second-class citizens. instead, designing around "all FEs are third party sites" forced the extension to have a real privilege model, to not expose long term key material to websites, ... to make it possible to write a first-class user experience as a third-party developer. long term, this is more important than the marginal security benefit of putting the FE in the extension. users who are concerned about FE/RPC privacy can easily run their own node and be immune to any of this (the full node can serve a FE on the RPC port).