Improve fwdctl dump

[rkuris]

Right now, the dump command only takes a database name. It would be easy to add options now:

• output formats json and/or csv (these would need to be streaming for large databases)
• specific starting key
• specific stop key
• maximum number of keys
• if we stopped, have an option to render the next key for resuming

Some of these stream limits are already implemented for range proofs, so you can use the code there as a template of how to stop iterating.

(depends on #347)

[rkuris]

import from json (or any format) would also be amazing