Fix destroy skipped on zero resource count when terraform state exists; reduce Run cyclomatic complexity

[Copilot]

When terraform apply fails mid-run, configuration.Status.Resources may be left at zero, causing the controller to unconditionally skip terraform destroy — orphaning any partially-created cloud resources indefinitely.

Changes

pkg/controller/configuration/delete.go

• Move the state-secret existence check before the Resources == 0 early-return. The destroy skip now requires both Resources == 0 and no state secret present. If state exists, destroy runs regardless of the resource count.
• Extract the three-part condition into shouldSkipDestroy(configuration, stateExists) bool for readability.

Resources == 0, no state  →  skip destroy   (nothing was ever created)
Resources == 0, state exists  →  run destroy  (failed apply left partial state)
Resources > 0 or nil, no state  →  run destroy + warn  (external backend: S3/GCS/etc.)

cmd/step/main.go

Resolved gocyclo lint failure (Run complexity: 33 → 13) by extracting five focused helpers:

Helper	Responsibility
waitForSignal	Block until wait-file appears or timeout/failure-file fires
runCommand	Single command attempt (pipe setup → start → wait)
runCommandWithRetries	Retry loop; returns (attempts, error)
handleCommandError	Touch error file, best-effort upload-on-error, return wrapped error
uploadSuccessFiles	Upload success-path files as Kubernetes secrets

Added uploadRetryAttempts = 2 constant to replace duplicate magic numbers across both upload paths.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.