Arranging different recycled cells to 10s5p pack

I'm collecting and testing used cells for my E-bike project for the past six months and during that time I was unable to find definitive answer on how to arrange cells of different capacities and different manufacturers in a 10s5p pack. My first thought was to sort the same manufacturer and same capacities within a parallel group (to spread load evenly), but that limits the total pack capacity to the lowest parallel group (times 10)

What I've pretty much confirmed using a few test cells with high difference between capacities (about 50%) in parallel is that mAh capacity pretty much adds up nicely, so I though it might be wiser to spread one manufacturer across all parallel groups, making each parallel group as random as possible, all while strictly taking care to equalize total capacities of all parallel groups. In such way, you minimize the risk of one manufacturer (from the same bad source) ruining the whole pack. On the other hand, you are intentionally mixing different cells with different discharge curves and different IR within a group

So far, my arrangement looks like this:



Different colors are different cell types. Measured mAh rating is writtern in a table cell. The first white line below represents how much is given parallel group capacity below or above ideal, average capacity. The second line is difference between strongest and weakest cell within a group.

Any thoughts?

Same amount of cells in each grup
Same amount of total capacity in each group
Same amount of cells per capacity range in each group (Lets say in 100mAh ranges)

This is the basis that ensures it will work

For higher current applications and or lower ps like yours you can consider
Same amount of model of cells in each pack
Same amount of cells with specific IR in each pack
Cutting out the ones with highest ir (Only on HIGH current packs) ( Can be ignored if testing current is done with same or higher than max current that will run later on)

Note that sorting by 1mAh per cell or anything under 100mAh is worthless when it comes to testers like Lii or Opus or equal cheap testers. This due to how much they can vary. But once again on lower p packs you have to sort based on smallest denominator. Best is to retest each pack afterwards when built.