I know it's a bit of an older thread... but... I don't agree with what some people are posting here. I think there's some misconceptions.
1 - The purpose of cell-level fuses is not to protect the cell if you tried to short the output, the way you typically think of a fuse. Most cells will be just fine if you try to abusively-quickly discharge them. Their internal resistance will spike and they will self-limit (somewhat) the amount of current that can flow out of them because the chemistry is limited at the rate at which it can act. They may get hot, but they'll soon be empty. They may be damaged by the heat and the higher-than-designed discharge, but that's okay.
In a many-parallel design, as soon as one cell gets hot, its resistance goes up, which means, relatively, it has lower voltage than the rest of the pack, so any current demands put on it by the load will be lightened on that cell and picked up by the rest of the pack. This also means that the entire pack is self-balancing load-wise. Cells in parallel will naturally share the load (and heat) equally.
It is electrically impossible for one cell to attempt to hog more of the load than others. That's what being in parallel means. Ditto for in series, they're forced to have the same current draw, that's Kirchoff's laws.
2 - That means that there's no point in having a cell-level fuse to protect each cell, only a single fuse for the entire pack. If they all share the load equally, then the only load you have to worry about being excessive is the total load on the output terminals of the battery. So yes, use one big fuse there.
3 - The purpose of cell-level fuses is to protect the entire pack from burning down if one cell fails. Most of the time when a cell fails (which happens somewhat randomly) it will fail "open", meaning the circuit through it is dead, like if you cut the ropes to a bridge, electricity will not get to the other side. Occasionally, rarely, when a cell fails it will fail "closed", meaning the circuit is shorted. If that happens, it means the entire energy of the entire parallel bundle is going to instantly discharge through that cell (which is basically equivalent to a wire across the + and - terminals). Instantly, that cell is going to reach ignition temperatures and liquify/blow up. The heat from which may set off its neighbors too in a chain reaction. So you have individual cells fuses to protect this. If a massive load is suddenly trying to discharge through that cell, something is wrong with that cell, and the fuse blows so that it fails "open". This removes one cell from the circuit, but otherwise has no impact.
...
So I don't get any of the logic which says that E-bikes and power tools should not have cell-level fuses.
Teslas have cell-level fuses, in high-current applications, that aren't rigid, with permanently mounted packs, that are massive numbers of cells. Pretty much shoots that argument in the foot.
However, the fuse only has to be thin enough to blow when the whole pack discharges through it, before that cell would overheat. Just about any connection, including nickle strips, will act as a fuse when an ungodly amount of current ramrods their way through it.
I wouldn't worry with actual fuses or delicate fuses. I don't know that I'd bother with anything smaller than 24g copper wire. Might as well just rely on the connections to the cells in most cases.