refactor: `ProcessedBlockProofs` and `FetchedDatalake` and `CompilationResult`

[rkdud007]

remove duplicated types. try to leverage already existing type if necessarily have to define new one.

plus, regarding two types there are lots of conversions happening inbetween HashSet <> Vec. intension to use HashSet was not loop multiple time if it's duplicated, but some case i'd need to turn into vector again and turn hashset again.

properly research about 1) how heavy is conversion 2) hows HashSet implemented underground so that could refactor the logic in optimal way