RetiredBA/BY, RR_NDB et al.
Albeit it might have come across as such I am not dismissing input from modeller airplane electrics - but there is one significant difference: I haven't seen many _paralleled_ LiPo's in RC aircraft. As I pilot RC aircraft as well I have the standard 1S, 2S, 3S and so on cells, but there are always n cells of 3.7V (nominal) _in series_. This is what the balancing circuit on the charger keeps an eye on - that the charge voltage stays equal at all cells.
The issue I seem to thnk about is more the paralleling of cells; as I mentioned before I consider unmatched cells (matching in terms of internal resistance, in 2nd approximation as well internal capacitance and inductance) not very suitable for high-demanding power applications. Whilst at relatively low current settings there are no issues, there certainly are some at high currents or changing loads (AC behaviour of a DC power cell). For as long as cells are just paralleled I agree with previous statements that the paralleling will provide rather stable DC characteristics. But then at AC conditions this looks certainly different.
Not to speak of the soldering quality of the connecting wires which in the past has not always convinced me of quality either. A minor non-perfect solder joint with some dirt in the solder tin can lead to funny effects, especially after some thermal cycling of the cell.
I fully agree with what has been said before, about the charging circuit not being ideal, about the AC conditions of a DC power source, about mechanical design issues when folding/winding up the cells, etc.
In the end, all these factors (and many others) are contributory to the risk and I wish these issues get solved.