Can we do away with the schema.json file?

This file is used to create an initial database, but it can pretty much be recreated at an point from the migration files. Without it, we might be better able to handle source merging. This would be because we may be able to replay individual merges out of order. I'd like to write some code that builds the schema file from migration files instead of reading it from disc. Then play around and see if we can get replays to replay migration scripts out of order.