Turn of the continueous mode fla

use the frame buffer? First call will allocate

Initialization for ST7735B screens

create a new step, apply its past waveform into the cyclic sample buffer and add a step_state object into active list so it can be added for the futur

returns interpolated value from delay buffer between int(findex) and int(findex+1) using the fractional part of findex.

Limited to 12 because we have 3 mixers funnelling into 1 mixer.

Merges the other VoiceGroup into this one, making additional voices available to this group.

Checking for new steps for pulse waveform has to deal with changing frequency and pulse width and not letting a pulse glitch out of existence as these

new steps for sawtooth are at 180 degree point, always falling.

Check for new steps using the phase update for the current sample for a square wave

TODO: This helps during refactoring, maybe it will be removed later.

MIDI-CC Only

Splits out num voices from the current group into a new group.

process all active steps for current sample, basically generating the waveform portion due only to steps square waves use this directly.

for pulse need to adjust the baseline according to the pulse width to cancel the DC component.

for sawtooth need to add in the slope and compensate for all the steps being one way

generate value for current sample from one active step, checking for the dc_offset adjustment at the end of the table.