Support for DB testing using aida-sdb

[HerbertJordan]

Currently, the main tool for evaluating Carmen is aida-vm-sdb which uses substate data and an EVM implementation to replay a sequence of operations on Carmen's StateDB implementations. This has two issues:

• it introduces the EVM processing overhead into the evaluation process, likely slowing down evaluation runs
• it introduces the EVM processing as part of the system-under-test, potentially complicating error diagnosis

The aida-sdb tool was originally conceived as a tool directly targeting the StateDB implementation itself, short-cutting the EVM implementation. However, its current status is not clear. In particular:

• it is unclear whether it is operational, and thus fit for practical use; it does not seem to be regularly tested in Jenkins
• it is unclear how to obtain traces for replaying operations; in particular, since it is not based on the Aida-DB trace information can not be obtained from this source
• it is unclear whether the trace format is compact enough to be practical; if traces take > 1 TB some test systems would require additional free disk space (e.g. profiling5)
• the tool has not yet been ported to the new executor infrastructure, thus lacking the latest features (e.g. progress tracking and hash validation)

The state of the tool and its potential should be evaluated and a decision should be made whether time should be invested in supporting this tool, or whether it should be removed.