Dry-run and validation for mutating operations

Summary

Agents should be able to preview the effect of mutating operations before executing them. A --dry-run flag on all mutating commands enables agents to validate actions and present them to users for confirmation.

Proposal

• Add --dry-run flag to all mutating operations (create, delete, start)
• Perform local validation before Kubernetes API calls
• Return would-be changes in structured JSON format
• Enable agents to preview actions before execution

Acceptance Criteria

• --dry-run flag available on all mutating commands
• Dry-run performs local validation and reports errors
• Output shows what would happen in structured JSON (when --output json is used)
• No actual Kubernetes API mutations occur during dry-run

Context

Part of #2849 — making tkn agent-friendly.

Note: Kubernetes already supports server-side dry-run (--dry-run=server). This may leverage that mechanism where applicable, but should also cover tkn-specific validation (e.g., parameter completeness for tkn pipeline start).

/kind feature

Start by locating the implementations and tests for the create, delete, and start commands, then trace how they validate inputs and call the Kubernetes API. Define the shared dry-run behavior and verify that local validation, structured JSON output, and zero mutating API calls meet the acceptance criteria.