Are All Three Necessary?

Not always. All image frames taken with a camera naturally include the bias. If you subtract a dark from the light frame, and subtract a flat-dark (one that matches the exposure duration of the flat) from the flat-field, then bias is already subtracted from both. You do not need a separate bias frame.

Bias frames are required if:

The simplest method is to use bias frames and auto-scale as required.  However, some sensors may work better with scaling than others.  So for the best quality we recommend using dark frames matching your light frames, and flat-dark frames matching your flat-field frames.

If the flat-field exposure is short, then dark current is negligible and you can just subtract the bias frame from the flat. You must as a minimum subtract either bias or dark from flat; otherwise the flat-field will not work properly.

Another strong recommendation is that you average the noise out of your calibration frames of all types.  You should capture at least 10 frames of each type.  The calibration tools in MaxIm DL will combine them automatically.

If you are not doing photometry (brightness measurement), and the image is sufficiently clean for your purposes, then you do not absolutely need a flat-field frame. Nor is a flat-field absolutely necessary for astrometry since the centroid algorithm used is quite robust; however it may help if the signal-to-noise ratio is poor. For general imaging, there are alternative methods to correct overall vignetting, such as the Flatten Background command. A can of compressed air is very effective for dust donuts. Nevertheless, if you want the best performance from your camera, or are doing photometric measurements, then a flat-field frame is mandatory.

As an absolute minimum, you should always subtract at least a dark frame. The camera control window includes an auto-dark feature that should be turned on if you are not otherwise calibrating your images. It is however always better to average dark frames if at all possible.