PurlDB-requested package index scans can fail for larger packages with large scans and many scans at once

[keshav-space]

send_scan_project_results in purldb-scan-queue-worker isn't able to properly send the large scan results.

This issue fixed the critical problems:

• The indexing of (very) large packages never completes purldb#362

And we can still improve this, using some of these approaches:

• Send compressed scan results: I was able to compress the ~350MB scan result to under 25MB using zstd or gzip ... xz is way too slow otherwise. For a start using the built in Gzip compression of HTTP clients and servers should help a lot.
• Adapt timeout to payload: The 60-second DEFAULT_TIMEOUT is not sufficient for sending scan results, we should have a variable timeout depending on the size of the scan results.
• Consider paginated upload in multiple chunks or streamed upload using JSON lines.