I know there is a lot of variability in the 1937/54 notes, even in the same consecutive runs! The only time I have been especially suspicious of trimming is if the note had considerable handling and/or folds yet has unduly sharp edges and corners for an otherwise inferior grade. However, I'm not sure weather or not a note's actually having been trimmed is all that important. I mean, the dimensions of the note are very important but the mechanism by which it achieved those dimensions is pretty much irrelevant.
Any loss of margin, even if the centering is improved or excessive mishandling is removed should reduce the value. So when the value of a small note is to be assessed, I think one ought to consider it on a worse case basis. That is, by imagining all the defects that could conceivably exist if only the margins were larger and evaluate accordingly.
Likewise, an oversize note should be considered Gem (or even a premium higher than) even if there is significant edge and corner handling, providing that this handling is confined to those areas that would not typically be part of the note. Even a significantly off-center note should trade at a premium provided the smallest sized margin is at least normal-sized.
I'm not sure how I would evaluate varying sizes of the printed design. This is much less noticeable than variations in the margin sizes, which is what I would be considering in reference to "size". Stretching or pressing might conceivably increase overall size, so perhaps only margin sizes should be considered at all.