Replace reindex API with bulk API and add multi-data-stream support

[AndersonQ]

When using the test kit some docs were failing because of how the reindex work and hiding the real issue. Thus I modified it to use the bulk as the Elastic Agent does. Also I made some small improvements:

Proposed commit message

Replace reindex API with bulk API and add multi-data-stream support

Switch from the server-side _reindex API to scan + _bulk with "create" actions, mimicking how Elastic Agent sends data. This gives per-document feedback: created, duplicate (409 version conflict), or other errors. Failed and duplicate documents are saved to separate NDJSON files.

Accept a list of data streams instead of a single one; the program now iterates over each data stream, scoping index names per data stream (e.g. tsdb-<data_stream>, tsdb-overwritten-<data_stream>).

Also:
- Read ES credentials from ELASTIC_PACKAGE_ELASTICSEARCH_* env vars with fallback to hardcoded defaults; mask passwords in output
- Fall back to system/certifi CA bundle when elasticsearch_ca_path is not set
- Replace exit(0) with sys.exit(1) for error paths
- Clear time_series_fields between runs to avoid stale state
- Update README with new configuration, examples, and output