Output paths: template artifact destinations safely

Use placeholders to structure outputs while respecting the sandbox.

Supported placeholders

Placeholder	Description
`{model}`	Model class name (or `combined` for aggregate files).
`{case_index}`	1-based index of the emitted shard or fixture case.
`{timestamp}`	UTC execution timestamp (defaults to `%Y%m%dT%H%M%S`). Supports `strftime` specs.

Examples:

pfg gen json models.py --include models.User \

  --n 3 --shard-size 1 \

  --out "artifacts/{model}/sample-{case_index}.json"

pfg gen fixtures models.py --include models.User \

  --out "tests/generated/{model}/fixtures-{timestamp:%Y%m%d}.py"

Placeholders resolve to segments restricted to [A-Za-z0-9._-]. Unsafe characters are replaced with _.
Paths that attempt to escape the working directory (for example ../) are rejected with an error.
Atomic writes ensure you either get the previous artifact or the newly completed file—never a partial.
Directories are created automatically; ensure your user has permission to write them under the project root.

Combine {model} with descriptive suffixes: tests/fixtures/{model}_fixture.py.
Use {timestamp:%Y%m%d} when you want dated snapshots while keeping deterministic content via --seed.
For parametrised fixtures, remember that case_index increments per case even when you limit models with --include.

See docs/emitters.md for emitter-specific behaviour and atomic IO guarantees.