refac: remove conditional compilation over DA types

[fakedev9999]

Removes conditional compilation over DA types by introducing WitnessData, WitnessExecutor, WitnessGenerator trait, ethereum and celestia modules.

[ratankaliani]

get_proof_stdin should not be async

[github-actions]

Performance Comparison

Range 2071356~2071361

Metric	Base Branch	Current PR	Diff (%)
Total Instructions	1704959414	1701943538	-0.18%
Oracle Verify Cycles	195682199	195682195	-0.00%
Derivation Cycles	1365411385	1362586288	-0.21%
Block Execution Cycles	7386315	7369762	-0.22%
Blob Verification Cycles	22996050	22745035	-1.09%
Total SP1 Gas	2401726496	2396873930	-0.20%
Cycles per Block	340991882	340388707	-0.18%
Cycles per Transaction	340991882	340388707	-0.18%
BN Pair Cycles	0	0	0.00%
BN Add Cycles	0	0	0.00%
BN Mul Cycles	0	0	0.00%
KZG Eval Cycles	0	0	0.00%
EC Recover Cycles	0	0	0.00%
P256 Verify Cycles	0	0	0.00%

[ratankaliani]

The recent changes to kona definitely simplified the diff. Have a few points which should be addressed before merging.

• Without a celestia feature on the proposer, how will the host be initialized with the CelestiaHost. IIUC you need the celestia feature on the proposer to activate the feature on op-succinct-proof-utils which activates the celestia host.
• The WitnessExecutor implementations can likely share more code.
• Is there a way to avoid the marker on the WitnessExecutor trait? That null pattern is typically an anti-pattern.
• Left some comments about propagating errors when possible using ?, rather than unwrapping.

[fakedev9999]

• Without a celestia feature on the proposer, how will the host be initialized with the CelestiaHost. IIUC you need the celestia feature on the proposer to activate the feature on op-succinct-proof-utils which activates the celestia host.

As replied above, no need to declare features of op-succinct-proof-utils. Passing in feature flag will enable feature in op-succinct-proof-utils.

• The WitnessExecutor implementations can likely share more code.

Not sure if this can be done without feature flags.

• Is there a way to avoid the marker on the WitnessExecutor trait? That null pattern is typically an anti-pattern.

The only alternative would be to store actual fields of these types, but there are two structs PreimageWitnessCollector and PreimageStore input as the oracle, which will end up having two structs implementing of WitnessExecutor with different types of oracle, which will lead to a lot of duplicated codes.

• Left some comments about propagating errors when possible using ?, rather than unwrapping.

Applied!