providers/infrastructure/v0.0.41

fix(infra): default empty source namespace in RuntimeNamespace (#469)

App Studio creates promoted Application/SimpleWebApp instances with no
namespace, so the instance object carries namespace "". kro still
materializes their children into "<cluster>-default", but the application
controller computed the secret-bridge target as RuntimeNamespace(tenant, "")
— a bare concat yielding "<cluster>-", which is both an invalid RFC 1123
label (trailing "-") and the wrong namespace.

The registry-pull-secret bridge runs before the fqdn stamp and returns on
error, so spec.expose.fqdn was never stamped and kro sat forever on
"no such key: fqdn (data pending)". Dev instances (no private image → no
registry secret → bridge skipped) stamped fine, which is why only promoted
prod instances hung.

Default an empty source namespace to "default" so the bridge targets the
namespace kro actually places the workloads in. Self-heals stuck instances
on redeploy.

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>