We split kv replay as follow:
- [LF] we introduce a simple protobuf to store submission entry (which
roughly correspond to TRANSACTION_ENTRY and PACKAGE_UPLOAD_ENTRY KV
submission)
- [KV] we create a tool to extract form a ledger export a sequence of
submission entry
- [LF] we move the replay tool from KV to LF, and base it on
submission entries instead of ledger export
The objectify of this split is double:
- on the one side it will simplify maintenance, as it will make API
between KV and LF more clear (depends only on a protobuf definition)
- on the other side, it will made the use of the tool handy, as it
will make possible to benchmark the engine without direct dependency
between KV and LF.
CHANGELOG_BEGIN
CHANGELOG_END
