Muggs, check out hownot2 vids where they go to a rope mfr and drop test, One vid is inside like an elevator shaft and another is an outdoor frame, there might also be a big indoor frame too. In these tests they often reach a puzzled conclusion "It shouldn't have broken" and they leave it at that. Squishy anchor point! Their breaks are on ridiculously rigid anchors like I beams.
On your tight zip ties on delta link it occurred to me that anything except perfect vertical centered drop can lever the delta link sideways against the outermost zip tie, focusing the load and breaking it first with a following series failure. In parallel load situations equalising the load is a challenge. It is sometimes achieved by having longer elements so the slightly shorter ones are able to stretch enough to let the load start to also settle on the longer elements. The longer the element the more give-distance it has to offer. In our case longer zip ties all set equally, hooked onto self-aligning biners might offer a better result. Seeing a good load distribution before shock loading and kind of holding that alignment with tape might help.
Yeah I'm winter inside bound.
The setup order of anchor load cell element lanyard weight is correct IMO. If it went anchor element loadcell Its wrong IMO because then the body of the load cell would have to accelerate to a new position in space as the element shock-load-stretched and this would add an f=ma error to the reading. post factum nerd alert
edit - I realise the irony that having a squishy anchor introduces f=ma error

An estimate is by symmetrical load cell design effectively 1/2 its mass creates error reading so eg 200 lb spike on 15 lbs 200/15=13G spike, applied to 1/2x1 lb load cell = 6.5 lbs error introduced to the reading.
edit2 - this issue is worthy of the fast-pull/slow-pull testing conundrum, with the extreme variant of fast-pull being shock-load. energy absorption and localised heating, or brittle fracture.