Hi PawPro,
First thanks for the compliment, I agree with your first impressions, it looks like a different beast (maybe not designed in-house??) to V3, but the waveform evidence so far looks as though it will yield to further study. The absence of a standard header of a series of '1's is not always necessary as any sequence in Manchester protocol should provide a steady signal bias around a middle level, as time spent at either On or Off will be equal. RFM69 certainly looks interesting, though I have not any experience at all with them.
Sorry I can't be of any further help, but I will watch your progress with great interest, so I hope you have a break through soon.
Rob