fix(runtime): substitute null for skipped steps in composite output.from

[scydas]

Summary

• A composite output.from like '{"a":${steps.x},"b":${steps.y}}' now
  substitutes null for references to skipped steps instead of discarding
  the entire aggregate into the step summary.
• Bare references (output.from: ${steps.x}) still degrade to the step
  summary when the step is skipped — no behavior change there.

Motivation: a pod-diagnosis workflow aggregates get-pod + list-events +
list-pods-by-node. The node-neighbors step is guarded with when (Pending
pods have no spec.nodeName). Without this fix, skipping neighbors discards
the pod and events data too — the user gets only {"status":"ok","steps":[...]}.

Test plan

• TestBuildWorkflows_CompositeOutputNullsSkippedStep — composite output with one skipped step produces null for that key, preserves others
• TestBuildWorkflows_BareOutputFromSkippedStepStillDegrades — bare reference still returns step summary (regression guard)
• Full test suite passes (go test ./...)

[codspeed-hq]

Merging this PR will improve performance by 13.53%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 8 improved benchmarks
✅ 13 untouched benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	small	193.6 µs	166.2 µs	+16.5%
⚡	large	2.4 ms	2.1 ms	+15%
⚡	BenchmarkFindCatalogCommand	8.7 µs	7.5 µs	+14.92%
⚡	BenchmarkCatalogJSON	2.9 ms	2.6 ms	+13.1%
⚡	large	1.6 ms	1.4 ms	+12.76%
⚡	small	144 µs	127.8 µs	+12.68%
⚡	json-small	254.6 µs	226.8 µs	+12.22%
⚡	yaml	5 ms	4.5 ms	+11.13%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing scydas:fix/workflow-output-null-skipped (5152660) with main (b3b0fae)}