You can "burn" the synthetic and claim part of the collateral.
I don't think there is a problem in this particular situation.
The place where there will be a problem is where MakerDAO had problems: when there aren't enough people willing to mint synthetics. The only reason to mint a synthetic is in order to sell it, in order to get a synthetic short position. If not enough people want to do this, but other people want a long position, this will generate excess demand for synthetics and the synthetics will trade consistently above the price of their real-world underlying.
This is precisely what happened to MakerDAO before it turned itself into a "stablecoin index fund". DAI was trading way above $1.00 for several months running. MakerDAO's goal was to provide a trustless stablecoin. If it costs $1.15 to buy one this week and $1.00 next week, it's not very stable. Those 15 cents didn't make DAI expensive -- they made DAI a failure at its goal of producing a stable coin.
Mirror's case is different. They're not trying to create a stablecoin. If the synthetics trade at a premium to the underlying, that difference is effectively a brokerage fee charged in order to open a long position. If the fee charged to open a long position swings around by 15% week-to-week that is certainly unattractive, but it doesn't make Mirror a failure.
This might actually work as expected. The likely failure mode will be annoyingly high fees for long positions, rather than failure to deliver on its promise (as happened with MakerDAO).
Why? The synthetic stock isn't convertible, so there's no arbitrage opportunity.