feat: add disk based event buffering

[mchristopher]

Adding support for buffering chain data to disk before uploading to S3. This is needed due to the growing data volume on chain, and the addition of the full input and output UTXO data now in place.

^{Cursor Bugbot found 2 potential issues for commit 64e6de2}

[cursor]

PR Summary

Medium Risk
Changes the live ingestion/upload path to use new disk-based buffering and streaming S3 uploads, which can affect correctness and recovery behavior under failures. Risk is mitigated by explicit cleanup/reset logic and added safety checks in multipart uploads.

Overview
Switches sv-dune from in-memory event accumulation to a disk-based Avro buffer (DiskBuffer) that flushes per block and finalizes into temporary Avro files for upload, reducing peak memory usage for large chain data.

Updates the service loop to buffer converted blocks/transactions/receipts to disk, then upload sequentially from file paths via Processor::process_data_from_file, resetting the buffer immediately after finalize to avoid infinite retry loops on upload failure.

Enhances Avro tooling with a global schema cache and a new AvroFileWriter, and extends S3 storage with store_from_file streaming uploads (including multipart-from-file) plus stricter multipart handling (abort on missing ETag). Also hardens runtime packaging by running the Docker image as non-root and adding basic debugging utilities.

^{Written by Cursor Bugbot for commit afe3f7b. This will update automatically on new commits. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}