Nobody needs to be able to code these in an interview. Ever. For certain domains you should be aware of them and be able to look up decent implementations. But to think that level of knowledge is important in an interview is bogus. I could just as easily ask similar questions and weed out most CS grads that get into Google or Facebook with these:

Please implement a first order low pass IIR filter. Tell me how the butterfly pattern in an FFT gets your from N^2 to N*logN. Oh, and implement an FFT. Write a basic PID controller implementation. Tell me how you'd handle a Field Oriented Control system that needs to run in voltage limit most of the time - what stability issues may occur? Write a fixed-point implementation of the sin(x) function. Implement a 2-pole 2-zero transfer function. For bonus points do it in fixed point without rollover or saturation problems. Assuming you have a matrix library available, give me the boilerplate code for a Kalman Filter. What kind of ODE solver should you use for long term stability when simulating planetary systems?

These are similar difficulty questions from a different domain, but many of them are likely to be used far more often in that domain than any of the interview questions in TFA are likely to be used in their domain.

The goal of an interview is to ascertain weather the candidate is capable of doing stuff and learning stuff, and if that's likely to carry over into the stuff you need done. It's not to see weather they can produce an answer to some specific problem on the spot. How you do that I'm not telling - it's hard enough without helping you find the people I need ;-)