Comments on FFT

[burdges]

Replaces: https://github.com/drahnr/rs-ec-perf/pull/2

We've largely understood the indexing by \omega_{j 2^{i-1}}, although not so well that we'd never miss any details, but issue that still require understanding:

• [ ] How are repetitions in skew factor being used? I suppose (19) or similar?
• [ ] (19) page 6286 in the proof of Lemma 4. Also some details of Lemma 2 I guess.
• [ ] Why skip that loop when skew != MODULUS not skew != 0?
• [ ] Layout of SKEW_FACTOR, like it only being field sized, not field sized times log of field size, probably (19) here too.
• [ ] Computation of SKEW_FACTOR

[drahnr]

Cherry picked the commit, closing.