add looping to rp2040 analogbufio.BufferedIn

[timchinowsky]

This PR contains part of the changes necessary to implement the processing of live sampled audio as described in #2676. Here analogbufio.BufferedIn is modified to add a loop keyword argument to BufferedIn.readinto(). If called without the keyword or with loop=False, behavior is identical to the current readinto(). If called with loop=True, readinto() sets up chained DMA which continuously writes ADC values into the buffer.

The contrast between the two modes can be seen by plotting the output of test code listed here:

import analogbufio
import array
import audiocore
import audiopwmio
import board

samples = 100
buffer = array.array("H", [0x0000] * samples)
adc = analogbufio.BufferedIn(board.A0, sample_rate=100000)

adc.readinto(buffer)

print("Sample,Print 1, Print 2,Print 3, Print 4")
for i in range(4):
    for j in range(samples):
        print(j, "," * (i + 1), buffer[j])

adc.readinto(buffer, loop=True)

print("Sample,Print 1, Print 2,Print 3, Print 4")
for i in range(4):
    for j in range(samples):
        print(j, "," * (i + 1), buffer[j])

Here is a plot of the data output by the first print loop when pin A0 is driven by a 1 kHz sine wave:

The buffer is plotted 4 times, but you can only see one (green) series, because when loop=False the buffer is static. Compare the plot generated from the second print loop:

The slight scatter between the series shows that the data in the buffer is different in each of the 4 print sequences. If the print loop is repeated again, the plot is similar, but has drifted in phase due to small differences between the clock rates of the ADC and the signal generator:

These plots are deceptively simple, because the signal frequency was chosen to put exactly one period of the signal in each buffer. If you change the frequency to 500 Hz instead the complications can be seen:

During the ~241 ms required to print the buffer 4 times, it has been refilled by DMA 241 times, alternating between the upward-going half of the sine wave and the downward-going half. Which half shows up at each printed sample will depend upon when exactly in the sequence that particular sample value was observed.

The other difference between loop=False and loop=True is that while readinto(buffer, loop=False) blocks until DMA is complete, then scales the ADC values to 16 bits before returning, readinto(buffer, loop=True) returns immediately after starting DMA, and the buffer values are raw unscaled ADC data.

[tannewt]

One super minor doc thing. Looks good otherwise. Thank you!

[tannewt]

Thank you!