Yes (usually temperature drift is a bigger issue), generally there is some (moderately complex) low pass filter that maintains a relatively stable baseline and recovers gracefully (no false positive, minimal false negative) from noise/moisture/thermal and other adverse events.
One interesting case is buttons in your pocket or face down on a desk. This is made more complex when low power requirements are made (e.g. lower sampling rate). A second is breathing on a very cold device. There are guarding methods that reduce the effect of moisture, but they may also run into IP challenges.
You cannot really maintain that. The controller has to adapt. That is what the two reset signals are for. The synchronous one can be trigger by application logic (for example some sort of timer) to re-calibrate the touch controller.
Intentional user input has fairly consistent characteristics, as do each of a variety of noise/drift/interference sources. The key is to safely and accurately recognize/estimate each of those to remove or report them. I suspect now people would prefer to use some form of machine learning, but realistically there are fairly low expected power requirements, which make that difficult. There's also a battle between low latency stateless techniques and more stateful techniques that require rapid recovery when they get something wrong.
