As I recall, the R/C folks match NiMh/NiCd cells to within 2%, or in the case of 1900mAh cells, 38mAh. I would think that with eneloops though, you're probably OK. I wouldn't go any more than about 100mAh though.
Another point, when matching cells for use in series, ideally, you need to discharge them on the C9000 at a similar rate as that which your speaker discharges the cells. For newer cells, this may not make much difference, but with older cells it may. As cells age, their discharge characteristics tend to vary more between cells at different discharge rates.
Also, to properly balance NiMh cells, you need to charge the "set" in series at a 0.1C rate for 16 hrs. This is the proper way to balance a dedicated NiMh, or Nicd pack. I don't usually do this though with "loose" NiMh cells, only welded packs.
I buy my NiMh AA cells in sets of 20. I then do a break-in with my C9000s and label the cells from, for example, 1-20, from lowest capacity to highest. Then about once a year, I do another break-in of all 20 cells, and relabel them, if necessary. For use in sets of 2-6 cells, this usually yields a "pack" that is typically within the 2% guideline that the R/C folks use. This is of course, provided I never match cells 19 and 20 with cells 1 and 2, for a 4s "pack" though.
Doing it this way, I negate the need for dedicated sets. I can just pull cells out, in numerical order, and they will be well matched for series applications. I'm not saying this is a better way to to do it than dedicated packs, but it does allow the original set of 20 cells to wear more evenly. I can verify this by noting that during a yearly break-in, the cells never switch position by more than one, or two places.
Anyway, I think you'll be OK with a 70mAh spread. I just wouldn't let it expand to more than about 100mAh. eneloops are pretty durable cells and I doubt that you would run into a reverse charging situation with the cells this closely matched anyway. Just remember though, closer is better.
Dave