Emit logs frequently for debugging

[costrouc]

Instead of only emitting logs at the end of a job completing/finishing instead we should emit logs at a set frequency. This will require the build to potentially be an async process to be able to periodically update the logs on file/object storage.

[costrouc]

This is an example of how logs are uploaded https://github.com/Quansight/conda-store/blob/main/conda-store-server/conda_store_server/build.py#L24-L31.

Right now this is only done on success/failure of the build. Instead we would like to be able to successively update/upload the logs.

https://github.com/Quansight/conda-store/blob/main/conda-store-server/conda_store_server/build.py#L147-L159 should be continually uploading logs at a set frequency that is configurable on CondaStore.log_update_frequency = ... [s].

[costrouc]

Redis would result long term in the best user experience.

1. check that redis is enabled
2. Potentially use redis streams https://redis.io/docs/data-types/streams-tutorial/ (but worth investigating if this is the right approach).

If a build is not completed/failed we can stream the logs via websockets, otherwise redirect the user to the storage for the logs (current behavior).

[costrouc]

Closing since frequent logs are currently emited. Please open another issue if more detailed logs are needed. It is hard to balance frequently writing logs and storing them on s3/object storage.