The idea would be to have a dedicated schema for each concept:
PM: path (subflows) creation and deletion (typically on the client side), and addresses announcements (typically on the server side, to tell it can be reached via another address, or one address of the client/server got removed).
Scheduler: where to send data on the available paths. (note: its goal might not be to use all the available bandwidth, it can also pick only one path (faster, less lossy, etc.), or duplicate data over multiple subflows)
The idea would be to have a dedicated schema for each concept:
You can get some inspirations from https://lpc.events/event/16/contributions/1354/attachments/999/1927/LPC2022%20-%20MPTCP%20-%20eBPF%20+%20Netlink.pdf (pages: 9 and 13)
While at it, it is important to mention that a "subflow" == "path" in the "Here is how it works" paragraph.