The processor with the display must always be fast enough to also handle RS-485 (unless you use another interface for this processor)

And if you must have a processor that is fast enough to handle the display, and fast enough to get the full state information for all slaves, then you have a hard time splitting "state master" from "display master". Unless the "state master" performs a lot of computations on the state before creating the state that should be displayed.

Where are your bottlenecks? Timing for RS-485? Amount of data to transfer? Computations on received state info? Foratting of graphics data? Emit of data to display? Are you sure that it isn't just a question of changing how the master is performing it's work? Can you offload more work to hardware? Can you just replace the master with another master with higher clock speed? Can you switch to a display with different electrical interface that may allow more of the communication to be handled by hardware? Can you change encoding of data, to send fewer bytes/message? Trade code or RAM space by precomputing information?

Adding more processors to a problem tends to increase cost and complexity, and doesn't always give the expected performance boost - sometimes the boost ends up being negative.