[WIP] [AI] Support target-image companions and complete static payloads

[TheSafo]

What does this PR do?

• Adds source_container: target to reuse the experiment target's already-pulled image for a container generator.
• Adds share_network_with: target to create the companion with Docker network mode container:target.
• Validates that registry and target image sources are mutually exclusive and that shared networking is not combined with disabled networking.
• Fixes static payload serialization so the complete file is transmitted byte-for-byte instead of only the first internal reader buffer.
• Adds configuration and property-test coverage for target companions and binary files larger than the old 8 KiB boundary.

Motivation

The logs-intake-backend regression detector needs to run a presigner companion from the target image while keeping it in a separate cgroup and sharing the target's localhost network. Its HTTP load also uses a binary zstd fixture larger than the previous static serializer's first-buffer limit.

This allows publishing an experimental Lading build for the Kafka-free intake CPU regression experiment.

Related issues

• Logs intake Regression Detector design

Additional Notes

Validation performed:

• ci/check
• ci/test -p lading-payload -p lading (490 tests)
• ci/fmt
• ci/validate passed formatting, checks, project lints, and Clippy; the repository-wide dependency audit then stopped on the existing RUSTSEC-2026-0204 advisory for crossbeam-epoch 0.9.18.

[datadog-official]

✨ Fix all issues with BitsAI

⚠️ Warnings

🚦 2 Pipeline jobs failed

Continuous integration | cargo-deny     

Security Audit | security_audit     

Useful? React with 👍 / 👎

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 2ba87ac | Docs | Datadog PR Page | Give us feedback!}