[doc] Cycles count info desired

[puzrin]

It would be very useful to know total cycles of fft run (with zero flash wait-states). That will help users to make estimates of CPU budget.

Examples of such info for CMSIS DSP:

• https://community.arm.com/cfs-file/__key/communityserver-blogs-components-weblogfiles/00-00-00-21-42/7563.ARM-white-paper-2D00-DSP-capabilities-of-Cortex_2D00_M4-and-Cortex_2D00_M7.pdf (page 10)
• https://community.st.com/s/question/0D50X00009XkhK9/fft-in-stm32f0 (reply 4)

This combinations: [ M0, M0+, M3, M4] * [ 1024, 512, 256, 128, 64 ] * [ cfft, cffti ]

Priority: after refactoring, not critical, not urgent