bdring/FluidNC v3.6.4-pre1

on GitHub

This is a test release for a problem that has affected several people - potentially everybody who uses RMT stepping. A change in the underlying ESP-IDF toolkit from Espressif led to an uninitialized variable that, in turn, cause the RMT hardware to randomly use the wrong clock. Sometimes it would work right, then after you reboot, it would be wrong and the step pulses would be too long by a huge margin.