One failing blob entry stops all blob downloads on the cloud until restart #85

Part of #80.

Core's blob downloader aborts on the first entry it fails to process, and the cloud never rebuilds it. After that, index entries keep arriving so new attachments show up in the data, but their bytes are never fetched and every request for them returns 404 until the process restarts. The only trace is a BlobStore error line on stdout. We have not seen a realistic trigger, so this may or may not be happening in the field; docker logs <cloud> 2>&1 | grep -c 'BlobStore error' on a deployment will say.

Fixes:

• fix(blob-store): skip entries that fail to process instead of stopping all downloads comapeo-core#1351 skips and logs a failing entry instead of stopping all downloads.
• The attachment audit (feat(attachments): add an attachment audit endpoint #88, run after each sync by feat(attachments): audit attachments after each sync and report missing ones to Sentry #90) surfaces the symptom on a deployment that does not yet have the core fix.