When download succeeds but convert fails, store provider source data
Description
In the fetcher pipeline, when the download job succeeds, and the following convert job fails, the whole pipeline fails. This results in losing the downloaded provider source data.
Actually it is not really lost: it is stored in the workspace of the pipeline on the server running the runner. But it is not pushed to the storage layer (Git repository) for source data.
What is desired: store downloaded data to the source data storage (Git repository).
The next pipeline should then run normally.
Caveat: the downloaded data of the first pipeline run won't be converted, except if the first pipeline was restarted manually after having fixed the convert script, with the option allowing to read source-data from the workspace of a previous pipeline run.