fix: recover failed setup resources before refresh

[alongubkin]

Summary

Direct setup refresh could re-enter InitialSetup while a setup-owned resource still had a failed status from the previous lifecycle. Initial setup then classified the resource as terminally failed before its controller had a chance to reconcile.

This change:

• adds lifecycle-filtered failed-resource retry to stack state
• retries only Frozen resources before a direct setup refresh
• leaves failed Live resources under runtime ownership

Tests

• filtered retry recovers a frozen refresh-failed resource while leaving a live failed resource unchanged
• failed setup lifecycle states remain eligible for direct setup preparation

Linear: ALIEN-629

[greptile-apps]

Greptile Summary

The PR recovers setup-owned resources before re-entering direct initial setup while preserving runtime ownership of Live resources.

• Adds lifecycle-filtered failed-resource retries to StackStateExt.
• Retries only Frozen resources during direct setup preparation.
• Adds coverage showing Frozen failures recover without changing Live failures.

Confidence Score: 5/5

The PR appears safe to merge, with no concrete correctness or security failures identified in the changed paths.

The filtered helper reuses established retry behavior, limits direct setup recovery to setup-owned Frozen resources, and leaves Live resource failures under runtime ownership as intended.

Important Files Changed

Filename	Overview
crates/alien-deploy-cli/src/commands/up.rs	Direct setup preparation now retries failed Frozen resources before entering InitialSetup while retaining the existing reconcile-and-release flow.
crates/alien-infra/src/core/state_utils.rs	Adds a lifecycle-filtered retry helper and a regression test confirming that Frozen failures recover while Live failures remain unchanged.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart LR
    A[Acquire deployment and reload manager state] --> B{Direct setup preparation required?}
    B -->|No| F[Run shared step loop]
    B -->|Yes| C[Prepare direct setup metadata]
    C --> D[Retry failed Frozen resources]
    D --> E[Set status to InitialSetup]
    E --> F
    F --> G[Final reconcile and release]

Loading

_{Reviews (1): Last reviewed commit: "fix: retry failed setup resources before..." | Re-trigger Greptile}