That is the obvious implementation workaround, and with such a 'canonical ID' software can do its own malleability-resistant transaction-tracking.
However, the hash over the malleable part is still protocol-significant: which exact incarnation of the isomorphic transaction is being passed around or cemented into blocks. So this new stable ID would be in addition to the older one, and might not even be necessarily expressed inside the protocol: it might just be a convention, and could vary across independent implementations.
The MTGox statement was a plea for the community to converge on such an consensus identifier before MtGox commits to a local fix. But that's not strictly technically necessary, so their stance looks like a strategy for blame-shifting and further delay. The Bitcoin core people don't like to rush into things.
However, the hash over the malleable part is still protocol-significant: which exact incarnation of the isomorphic transaction is being passed around or cemented into blocks. So this new stable ID would be in addition to the older one, and might not even be necessarily expressed inside the protocol: it might just be a convention, and could vary across independent implementations.
The MTGox statement was a plea for the community to converge on such an consensus identifier before MtGox commits to a local fix. But that's not strictly technically necessary, so their stance looks like a strategy for blame-shifting and further delay. The Bitcoin core people don't like to rush into things.