The RTL8192C (at least) firmware rate control behaviour seems to be
trying the highest configured rate first, and then trying slower rates
over time. It doesn't SEEM to be tracking it inside a given macid in
firmware - so if your rate mask says MCS15 is possible, but you're not
able to successfully transmit at it, the firmware will still keep
trying that every frame.
The vendor and Linux drivers work around this by looking at the RSSI
reports and constraining / expanding the ratemask for each macid.
This just adds the placeholder config field and rate masks; later
commits will set those and use those appropriately.