Fix random failure: MissingScriptWitnessesUTXOW (fromList [])

[TimSheard]

For a long time we get random, in frequent failures in the property tests labeled with
MissingScriptWitnessesUTXOW (fromList [])
This is annoying, and actually very misleading because the actual failure is that there is an extra script witness that is not actually needed. This misleading error message is caused by a long known bug that was improved by code like this

validateMissingScripts pp sNeeded sReceived =
  if HardForks.missingScriptsSymmetricDifference pp
    then
      sequenceA_
        [ failureUnless (sNeeded `Set.isSubsetOf` sReceived) $
            MissingScriptWitnessesUTXOW (sNeeded `Set.difference` sReceived),
          failureUnless (sReceived `Set.isSubsetOf` sNeeded) $
            ExtraneousScriptWitnessesUTXOW (sReceived `Set.difference` sNeeded)
        ]
    else
      failureUnless (sNeeded == sReceived) $
        MissingScriptWitnessesUTXOW (sNeeded `Set.difference` sReceived) 

Note when (HardForks.missingScriptsSymmetricDifference pp) is False, and sRecieved is a supersetof sNeeded, we get the misleading error.

So the real problem is that the EraGen process of generating random traces, occasionally (and randomly) generates an extra script witness. It is hard to track this down. There is some evidence that this arises because the script hash appears as the PolicyId in a Value. But that is not been verified. Regardless of the cause, we fix this by
discarding the trace. Note that this is failure to generate a "random" but valid transaction. Discarding the trace adjust for this inadequacy in the generation process. This only appears in the Alonzo era, so this "fix" is applied here, in the genEraDone
method of the EraGen class in the (AlonzoEra c) instance.

[JaredCorduan]

I believe this PR will resolve #2839 and #2811 🎉

[lehins]

Thank you for finally fixing this annoyance

[lehins]

👍