insights/compute: improve compute client to expose skipped reasons

Created by: leonore

this is an issue parallel to https://github.com/sourcegraph/sourcegraph/issues/36226 for compute.

we'd like the compute client to expose skipped reasons so that we can retry on shard timeouts.

according to the search team:

I think you want something like progressAggregator for compute. It will be updated with the compute Event type, and you’ll probably need to add the Stats field (which is how we stream repo status) to the Event result similar to how it’s done on the SearchEvent type. Then, you’ll just want to either add a progress event to the output stream (similar to how it’s done in the search stream), or since you probably don’t care about getting progress updates incrementally, you could probably just add that information to the done event.

once that is done we can then use those skipped reasons in the insights decoder like it will be done in the issue above.