I am developing an INDI driver for a mount we're building at work, and I'm fairly new to astrophotography concepts. After looking through the drivers which use INDI::GuiderInterface, I am unclear on a few points I was hoping somebody could help me with please?

I think I get the basic premise of how manual guiding would work on a mount with an ST4 interface. You set your guide rate, and then while holding the button down on the paddle a GPIO signal tells the mount to increase the rate that the corresponding axis is rotating by that amount. My understanding is that a CCD with an ST4 output does the same thing, by measuring the drift of the image and calculating a pulse duration needed to keep it centered.

My questions are about how INDI::GuiderInterface ties in to all of that:

1. How does the driver know which slew rate to use for calculating pulse durations? Or will it just pick whichever one I have active in device manager?
2. If it does just use whichever slew rate is active, how does the client (ekos or PHD2) know what that rate is? The Telescope base class doesn't seem to have a getSlewRate() method.
3. Am I correct that the driver should interpret the pulse as "tell the RA/Dec axis to go this much faster than it is already going for this many ms", so for example if my active slew rate is 2x sidereal and the client wants it to speed up the RA axis, it will have the effect of going 1x+2x=3x sidereal in the RA axis?
4. Is the guide interface active all the time? Looking at the Paramount driver for example, it only seems to care if the mount is parked or not when the guide interface methods are called. Is there some mechanism in the client that decides to only make calls to the guider interface when the image is more or less stable, or is it based on the state of one or more properties in the driver?
5. Finally, does the client use the whole image to decide if a guide pulse is needed, or is it based on a sync point? In other words, is it considering a specific object in the image (based on a plate solve or location on the CCD or something), or just whether or not the image as a whole is drifting?