Next step after fixing misconception in ForkedChain

[jangko]

after we fix #2949

the next step after that is going to be to add to BlockDesc a layer or blockT field that holds the state associated with having applied that block - any time a fork is created, instead of replaying via stagingTx, this field will be used instead as a starting point - I've mentioned a few other, more efficient designs in the past, but this is the most simple one that gets us over the finish line