My responses to your questions are below. The attached document contains these responses as well as additional notes.
I agree that when you have a series of numbers some of which are paired with a sign of operation (e.g., "5, 8, 2, -3, 6, and -5"), it is preferable to enclose the whole series of numbers in one set of Nemeth switch indicators. However, if only two numbers appear, one of which is paired with a sign of operation, then I would enlose in Nemeth switch indicators only the modified number.
Author's notes within spatial problems ... should likely be handled as you describe, keying each author's note (using a numbered key) and listing the notes before the equation. As for the begin Nemeth Code indicator before spatial material when a non-embedded transcriber's note precedes the spatial material, I would not place the indicator on the line with the end of the TN, because the Guidance only technically allows for the indicator to be placed after preceding text. I think the likelihood that the reader would miss the opening Nemeth Code indicator is too great if it is placed following a TN.
For a multi-word author's note that appears with a fraction with cancellations (which must be represented spatially), I would do as you have suggested and use the keying technique described in item #2 above.
Could author's comments be left uncontracted even though they are not technically part of the Nemeth equation? No; that would be unclear and inconsistent. Author's non-mathematical comments should be in UEB.
Thank you for your questions and for your attention.