The "basic math" doesn't pencil out. Air France flies an A350-900, same plane as Delta's A350-900, with Safran Optima seats, the same seat as United's Polaris. Air France fits 34 Optima seats in between doors 1 and 2 along with half a galley, while Delta manages to fit 32 seats in between doors 1 and 2 with no galley and the center row 9 seats going into the space between doors 2. You can definitely cram in 1 more row of Optima seats and match the amount of space that D1 is taking up, maybe even cram in a second row in the center, so that would result in 38-40 Optima seats taking up the same space as 32 D1 seats, or 19-25% more seats vs D1. If the difference in the number of J seats is due solely to density, a 40J "high-J" (for a handful of the most premium routes) D1 cabin implies a 48-50 seat Polaris cabin for premium routes on an A350-900 sized plane. Does that match up with how UA has configured the planes that it uses for premium routes (let alone the most premium)? Not at all. I would argue that the 787-9, 777-200 and 777-300ER are most comparable to the regular 32J D1 as UA flies a large fleet of each of those aircraft on a variety of (mostly premium long-haul) routes, while the high-J 767 is a a small sub-fleet serving the specific purpose of putting in more J capacity for a handful of routes with outsized premium demand, while the new high-J A350 is (at least from indications so far) a small sub-fleet serving the specific purpose of making the South African extra-long routes feasible by cutting down passenger count.

UA's 787-9, a plane with ~10% less maximum passenger capacity, has 48J vs an implied ~43-45J based solely on density and plane size differences from the new Delta A350 D1 configuration and ~34-36J from the old configuration.
UA's 777-200 has 50J and the plane has the same capacity as an A350-900, implying 48-50J from the new D1 configuration and 38-40J from the old D1 configuration (UA uses this for EWR-HND and IAD-HND while DL uses the standard A350 for the comparable ATL-HND and DTW-HND routes).
UA's 777-300ER has 60J and the plane has ~20% more capacity than an A350-900, implying 57-60J from the new D1 configuration and 46-48J from the old D1 configuration.
UA's high-J 767-300ER has 46J but the plane only has ~2/3 the max capacity of an A350-900, which implies the Polaris cabin should be ~31-33J based on the new D1 configuration or ~24-26J based on the old D1 configuration (and even the regular UA 767-300 has 30J).

Looking at those stats, it's clear that UA configures its standard fleets with enough Polaris seats to match or exceed the top end of what the high-J A350 D1 seat count would imply when adjusting for density and plane size, and with significantly more Polaris seats than what DL's regular A350 fleet D1 seat count would imply. How do you explain the rest of the difference?

Beyond the "basic math" that clearly shows UA seeing more J demand, J seats are not priced on a square footage basis. A cursory check of markets where UA and DL directly compete such as LAX-LHR/SYD/AKL/HND and EWR/JFK-LHR/MAD/BCN/DUB/MXP/FCO/GRU (ignoring the dominant CDG/AMS Skyteam hubs and FRA/MUC/ZRH Star Alliance hubs since UA won't have enough frequencies to be competitive on CDG/AMS and vice versa for DL) would tell you that there is no consistent D1 premium vs Polaris proportional to Polaris' smaller square footage. All of that would tell you that UA put in more J seats not solely because they could fit more Polaris seats into the same space as D1, but that they see more customer demand for J in its route network than DL does.