Raft Consensus with a Minority of Nodes

A nice observation! As far as I know, generalizing from majority to specific constraints on quorum intersection was first proposed in Heidi Howard's April 2019 PhD thesis, Distributed consensus revised. In chapter 4, Quorum intersection revised [pdf], she outlines two refinements to Paxos quorums; in the first:

Our first finding is that it is only necessary for phase one quorums (Q1) and phase two quorums (Q2) to intersect. There is no need to require that phase one quorums intersect with each other nor that phase two quorums intersect with each other.

And the second, probably more similar to the one given here (slightly edited to be readable in non-PDF form):

We require only that a phase one quorum of the epoch E intersects with the phase two quorums Q for all smaller epochs. There is no requirement that the acceptors for the phase one and phase two quorums for a given epoch intersect. Likewise, there is no requirement that a phase one quorum of the epoch must intersect with the phase two quorums for all greater epochs.

No shade to the author intended, I don't think they claim novelty here.