Store one dataset CSV per annotator session

Add resume-from-dataset flow after refresh

Update annotation datasets and summary

Add durable dataset backup for annotation responses

Initial FinPersona Docker Space

initial commit