GTFS schedule rewrite: Implement task result base class

[lauriemerrell]

We have a ProcessingOutcome base class, but no JobResult base class. This results in us reimplementing very similar versions of the following code in every since job/task (specifically: the schedule download job, schedule validation, RT parsing, RT validation, schedule unzipping, and schedule conversion to JSON.)

We should add a base class with this and then just subclass it in each instance to avoid copy/pasted code.

class JobResult(PartitionedGCSArtifact):
    bucket: ClassVar[str] 
    table: ClassVar[str]
    partition_names: ClassVar[List[str]]
    outcomes: List[ProcessingOutcome]

    @property
    def successes(self) -> List[ProcessingOutcome]:
        return [outcome for outcome in self.outcomes if outcome.success]

    @property
    def failures(self) -> List[ProcessingOutcome]:
        return [outcome for outcome in self.outcomes if not outcome.success]

    def save(self, fs):
        self.save_content(
            fs=fs,
            content="\n".join(o.json() for o in self.outcomes).encode(),
            exclude={"outcomes"},
        )

[lauriemerrell]

We think this may be lower priority than moving on to warehouse work.

[atvaccaro]

Currently we've been using .jsonl.gz for data and .jsonl for outcomes. We could continue this pattern and enforce it in the type; it makes it mildly easier to look at outcomes which really shouldn't be that large anyways.

[lauriemerrell]

Maybe revisit this when doing Payments work.