Support baselines in runs and make baseline_id optional

[dnerini]

Motivation

• Allow baseline entries to be defined inline in the runs list (alongside forecasters and interpolators) and simplify baseline configuration by making baseline_id optional.
• Preserve backward compatibility with the legacy top-level baselines block while encouraging the new inline format.

Description

• Update the Pydantic model so BaselineConfig.baseline_id is optional (str | None) and its description documents that it defaults to the label when omitted.
• Extend ConfigModel.runs to accept BaselineItem entries and mark the top-level baselines list as deprecated with a default empty list.
• Update the JSON schema (workflow/tools/config.schema.json) to reflect the nullable baseline_id, allow BaselineItem in runs, and remove the requirement for the legacy top-level baselines field.
• Modify Snakemake helper workflow/rules/common.smk so baseline entries in config['runs'] are ignored by collect_all_runs, collected by collect_all_baselines (using baseline_id fallback to label), and legacy top-level baselines remain supported.
• Adjust example configs to remove the old top-level baselines block where applicable and add/maintain inline baseline entries.
• Add unit tests to validate backward compatibility and correct workflow parsing: test_legacy_top_level_baselines_still_supported and test_workflow_parsing_excludes_baselines_from_run_configs.

Testing

• Ran the updated unit tests (pytest tests/unit/test_config.py) focusing on config validation and workflow parsing, and all tests passed.
• Verified the two new tests test_legacy_top_level_baselines_still_supported and test_workflow_parsing_excludes_baselines_from_run_configs succeed.

---

Codex Task