cloud calibration updates

[cdc-as81]

Summary

Adds cloud and Docker-backed phase 1 calibration support for ixa-epi-covid, including reusable MRP runners, Azure Batch orchestration, synthetic population staging, auto-sizing, cleanup tooling, and expanded test coverage.

Changes

• Add packaged phase 1 calibration entrypoints with direct, local MRP, Docker MRP, and Azure/cloud execution modes.
• Add Dockerfile.cloud and MRP config files for local, Docker, cloud, and cloud-task execution.
• Add cloud runner support for:
  • staging one shared synthetic population file per cloud session
  • rewriting task payloads for container paths
  • reusing Azure service clients
  • limiting parallel output downloads
  • optional RAM-based cloud auto-sizing
• Add Azure cleanup CLI for listing and deleting cloud session resources.
• Refactor model execution into shared helpers for importation generation, IXA execution, output reading, and canonical output.csv writing.
• Update Makefile targets for cloud sync, Docker image builds, phase 1 calibration modes, checks, and help output.
• Improve synthetic population generation by deriving household frames from sampled PUMS data and filtering invalid household weights.
• Update FIPS/setting code capacity to support larger setting IDs.
• Prefer packaged importation data when available, with cache/download fallback.
• Add tests for cloud runner behavior, population asset staging, auto-sizing, MRP runners, cleanup wiring, phase 1 helpers, and model execution contracts.

[KOVALW]

Commenting here that we'll keep this PR as a draft until calibrationtools has a front end that doesn't require more than a TOML and minimal argument changes to the sampler run calls in order to use Azure cloud