fix(workflows): fail fan-in loudly on a non-string wait_for entry

[Noor-ul-ain001]

Summary

FanInStep.execute guards a non-list wait_for (added in #3482), and the engine's load-time validation (engine.py) rejects non-string wait_for entries. But the engine does not auto-validate step config before running it — the same "unvalidated run" scenario the existing non-list guard is written for — so execute iterated the list's elements raw and hit two failure modes:

• Unhashable entry — a list/dict from a natural YAML indentation slip like wait_for: [[a, b]] crashed the whole run at context.steps.get(entry, ...) with a raw TypeError: cannot use 'list' as a dict key (unhashable type: 'list').
• Hashable non-string entry — wait_for: [123] silently joined an empty {} and still reported COMPLETED — the exact "silent empty result + COMPLETED" wiring bug that the whole-list guard and the engine's fan-in validation both exist to prevent.

This extends the execute() guard to reject any non-string wait_for entry, mirroring the sibling non-list guard directly above it and reusing the engine's own "entries must be step-id strings" phrasing for a consistent error.

Reproduction (against current main)

from specify_cli.workflows.steps.fan_in import FanInStep
from specify_cli.workflows.base import StepContext

step = FanInStep()
ctx = StepContext(steps={"a": {"output": {"x": 1}}})

# 1) unhashable entry -> raw TypeError takes down the run
step.execute({"id": "j", "wait_for": [["a", "b"]]}, ctx)
# TypeError: cannot use 'list' as a dict key (unhashable type: 'list')

# 2) non-string entry -> silent empty join, reported COMPLETED
r = step.execute({"id": "j", "wait_for": ["a", 123]}, ctx)
# r.status == COMPLETED, r.output["results"] == [{'x': 1}, {}]   <-- bogus {}

Changes

• src/specify_cli/workflows/steps/fan_in/__init__.py — reject any non-string wait_for entry in execute(), failing the step cleanly with output={"results": []}.
• tests/test_workflows.py — parametrized regression test (test_execute_non_string_wait_for_entry_fails_loudly) covering unhashable (list/dict) and hashable-non-string (int/None) entries.

Testing

• New test fails on the pre-fix source (raw TypeError for unhashable, wrong COMPLETED for non-string) and passes with the guard ("test-the-test" verified via git stash).
• All fan-in tests pass (23 selected).
• Pre-existing unrelated failures: TestWorkflowCliAlignment symlink/cross-project tests fail on Windows without Developer Mode/elevation — verified identical on a clean upstream/main tree with these changes stashed.

Related

Same "unvalidated execute() crash" class as the sibling fan-in non-list guard (#3482), fan-out non-list items/step-template guards, the if/switch non-list branch guards (#3515), and the while/do-while non-list steps guards (#3519).

🤖 Generated with Claude Code