Using input C to manage stability.....not sure I would rely on that in a production
environment. T affects on board parasitic C ......and fdbk R drift/tolerance, one
would have to run the numbers.
The OP schematic is not identical to a conventional sallen-key. RB is not current-carrying in a significant way. Changing RB does not affect rolloff curve in simulation anyway.
Usually we take an input from some sort of divider network (resistive, or RC, etc.) In fact simply by lifting the right end of C1 and attaching it instead to the inverting input. That makes the schematic behave. It's intended to be a high-pass filter, correct?
Rb is used for the OPs without built-in bias current cancellation, there's no other useful purpose for it, I think. The resistor value is most likely a typo, respectively R1 and R2 have been modified without thinking about Rb.
LT2050 is apparently also a typo, the type doesn't exist.