What about decreasing the margin to 2x and using fees to recapitalize losses from blown shorts gradually over time? That way, XTS holders who don't touch BitUSD don't risk a sudden dilution of their capital from an adverse event. Rather, dividends are reduced or eliminated.
You could also "buffer" the fees in a reserve balance, which would simplify bookkeeping enormously. E.g., whenever a blown short happens, you just add the negative BitUSD balance and positive collateral XTS balance to the reserve balance, and zero out the position attached to the account like a normal margin call. Then the total negative BitUSD in the reserve balance is how many BitUSD you want to purchase on the market to get rid of the excess uncollateralized BitUSD, and the positive XTS balance is how much capital you have to make that happen. The actual open market operations to be performed with the reserve balance would be specified by some fixed algorithm.
The reserve BTS balance should decay exponentially over time (I'd suggest a half-life of ~120 days). This means that reserve balance will eventually be paid as dividends to XTS holders (by destroying XTS) if no blown shorts occur. Also, the fraction of the reserve available for open market purchases in a single block, and the market depth to which it can go, should probably be limited.
In an event with lots of blown shorts, the reserve funds might be inadequate to make good on them right away, but BitUSD holders will know that eventually the extra money will be destroyed.
Of course there are lots of flourishes you could add. For example, the reserve BTS balance shouldn't decay when the reserve is in danger of insolvency. (I.e., XTS assets of reserve balance divided by XTS liabilities of reserve balance gets lower than some threshold K, e.g. K ~ 2, as determined by valuing BitAsset balances at the current market price). Or maybe we could temporarily increase fees when the reserve balance is hurting.