I don't think the quality of prime RC cells is as bad as you say. I do avoid the cheapest packs and move a little up the quality ladder. Buying the absolute lowest priced product is generally not a good QC plan. But once you get off the "bottom" of the market the quality has been quite good, and it has improved markedly over time as they learn how to make better product.
The failure rate is nothing like 10%. A pack has 6 or so cells, so a shipment of 13 packs has almost 80 cells, and zero were bad in my CroBorg's Multistar order, as an example. The out of the box balance voltages were much better matched than I see on new prime A123 cells, for another example.
My average orders are 50-100 cells, and there have been less than one bad cell per order on average. So my lipo bad cell rate has been well under 1%. People who mention numbers like 9 of 10 are not being precise. It is just part of the philosophy to buy a few extra cells to insure a complete pack's worth of good cells are available during the build in the rare case that some are bad.
Every ebike shop that has a welder is making packs, and increasingly home enthusiasts are doing it too. Are those the same quality as commercial factory quality?
Just like the cheapest lipo, there is the cheapest 18650. I've had those burst and corrode in flashlights, and have poor life. Good or Bad Quality is not reserved for any particular cell type.
I think the problem here is "low quality homebuilt (or commercially built) packs", not "lipo". Lipo is attractive for many reasons, but dangerous packs are built out of many cell types. Headway LiFePO4 packs with a BMS have burned just fine as well.
If lipos are inherently dangerous how can they be used in so many commercial products?
For lipos to be safe in cellphones and tablets requires electronics that works correctly. Laptop series parallel packs also requires balancing that works properly. Clearly the electronics can be done correctly, or these phones and laptops would be burning up regularly, like they used to.tor
Manual Monitoring various lipo packs that I have built, I see they require balancing less than once per year with hundreds of cycles per year. I'd like to have a reliable BMS but the high vibration environment of eBikes is hard on electronics, and BMS failures are common. The experience for many has been that BMS's kill more packs than manual monitoring.
We need better quality packs and better quality BMS's at prices sensible for the eBike market. The low quality BMS's on most eBike packs don't meet commercial BMS standards for redundant self checking. A BMS that doesn't self check is not really any better than manual testing. Most failure modes result in the BMS indicating all OK. If it fails you think you are protected, but you are not.
As others have mentioned, many failures come from shorts. These are irrespective of cell type.
Many failures come from charging a damaged pack. In many or most cases the operator knows the pack is damaged and charges it anyway, hoping it will recover. A functioning BMS or a proper manual check would prevent this. A broken BMS may provide a false sense of security and lead to a fire.
There are differences between cell types and between chemistries. But all the other important factors cannot be ignored. Choosing one factor and ignoring all others doesn't provide a balanced view of the risks.
Keeping safe requires understanding the factors, making considered plans and taking appropriate actions, not fixating on one issue and ignoring other more important ones.
Stay safe on this fourth of July weekend, and have a good one. We want our fireworks to burn safely and our ebikes to avoid burning.