2) What is the best general glass MXM for interior/exterior use? Needs to have reflections, refractions and render reasonably quickly when viewed internally or externally. Or should I be using a unique glass MXM for different situations?
If you don't necessarily need refractions (for a thin glass window it's in many cases not necessary) you can use 'AGS' instead which gives you reflections but doesn't refract the light. Do a search for AGS, it will become pretty clear what it does and why you should use it to save some render time.
4) what is more efficient in rendertimes for displacement - more geometric detail or higher precision value?
More detail in the original geometry. This way you don't have to raise the precision value very much. Maxwell can deal with many polygons without a problem so always best to subdivide a bit the geometry you intend to use displacement on.
5) what is the best file format to export an MXI to to correct white balance? EXR? HDR? What is the best software to handle this in post?
HDR will do fine, I guess any 2D editing software you're comfortable with. Photoshop has some plugins that make this easy for you, I suppose also others.
8) does it always generally take a long time for emitter light to appear in reflective materials? (such as a light reflecting in a pool of water).
Depends on the scene and how complex the lighting is, usually reflections of emitters appear pretty fast. Maybe you mean caustic light seen through a refractive material?
10) when rendering a COOP image, is it best to set up the COOP first, or is it ok to just render the same image on 2 machines and merge the two, even if they covered the same SL range? e.g. both machines render an image from SL0 to SL12, then merge, does this produce SL24 or just a less noisy SL12? Or do I need to render the image on one machine from SL0 to SL12, then render on the other machine from SL12 to SL24 and merge?
Whether you set up a coop render using the Network rendering, or do it yourself manually and then manually merging the MXIs, it's the same result. Coop render means several computers work on the same image at once, so if one computer took 10 hours to render to SL 18, then 2 computers will take around 5 hours to render to SL 18 the same scene. When merged, the SL won't double, it will be the combined SL reached depending on how many MXIs you merge. As the SL increases so does the time to reach the next SL, almost exponentially. If it took 30 seconds to go from SL 1 to 2, it would be great if it also took 30 seconds to go from SL 16 to 17....
11) apart from emitters, what are the common features of a scene that can slow down a render? displacement? reflectivity of materials? physical sun? do they make much of a difference as with other renderers? any general rules of thumb for keeping rendertimes down? (apart from keeping Ref 0 RGB <240, using small and lowpoly emitters etc.)
Use AGS for windows instead of real glass if you don't need refractions
Displacement will slow down a render, it depends mostly on the precision setting and secondly the height. Larger displacement>more render time, but precision is the most important setting, that's why it's good to have a pretty subdivided initial geometry.
Try to keep the scene reasonably within the bounds of what the camera sees, ie if you're trying to render one house in a neighbourhood, don't build your scene consisting of 50 houses that spans several kilometers. It will add to the render time because Maxwell has to calculate the lighting for everything. So if you think something in the scene is too far away to have any influence on the lighting of the portion of the scene that your camera sees, just hide or delete it from the scene.