An IIR filter will always beat an FFT, but the most efficient way to implement longer FIR kernels is often with a WOLA or similar "streaming" FFT filter implementation. At some point, it takes fewer cycles to do N*log(N) multiplication in the frequency domain than to do N^2 multiplications to convolve in the time domain.

The crossover point will be system-dependent and heavily influenced by overhead, but a crude WAG might be in the vicinity of 64- to 128-wide kernels. There is no question of one implementation being "better" than the other, they are capable of identical results if implemented accordingly.

Applying a FIR in frequency domain faster for 1D filters longer than some very small number, usually 7-16. The old school graphics and audio filters are shorter sure, but they are also horribly bad compared to what you can do with a designed filter in the 20-40 range.

Some standards define a frequency response of the whole measurement chain, including things like the frequency response of your sensor and anti aliasing filter. An FFT is the easiest way to deal with that.

When you want to do signal analysis in the frequency domain?