feat(ci): Add HA support for node failures

Modified CI/CD workflows to gracefully handle Proxmox node failures:

CI Workflow (ci.yml):
- Terraform Plan only runs on main branch (faster CI on feature branches)
- Plan failures on unavailable nodes don't block validation
- Added warning message when plan fails

Deploy Workflow (deploy.yml):
- Added continue-on-error to all deploy jobs (pve1, pve2, pve3)
- Modified cluster validation to require 2/3 nodes (quorum)
- Enhanced deployment summary with success counter
- Exit codes: 0 if >=2 nodes, 1 if 1 node, 1 if 0 nodes

This ensures the infrastructure remains operational even when one
Proxmox node is down, maintaining HA principles.

.forgejo/workflows/ci.yml

@@ -35,7 +35,7 @@ jobs:
           done
 
       - name: Terraform Plan
-        if: github.event_name == 'push'
+        if: github.event_name == 'push' && github.ref == 'refs/heads/main'
         run: |
           for dir in terraform/pve*; do
             if [ -d "$dir" ]; then
@@ -44,7 +44,7 @@ jobs:
                 cd "$dir" && \
                 cp ../terraform.tfvars.example terraform.tfvars && \
                 tofu init && \
-                tofu plan -out="tfplan-$(basename $dir)"
+                tofu plan -out="tfplan-$(basename $dir)" || echo "WARNING: Plan failed for $(basename $dir) - node may be unavailable"
               )
             fi
           done