[FEATURES] Support creating Runtime without Dataset dependency

[osswangxining]

What feature you'd like to add:

Background

In some corner cases, users need to create Runtime without pre-defining a Dataset resource. Current Fluid architecture requires strict Dataset-Runtime binding, which brings unnecessary complexity for temporary data scenarios. we propose allowing creation of runtime without Dataset dependency through Kubernetes annotations.

### Annotation Design
```yaml
apiVersion: data.fluid.io/v1alpha1
kind: JuiceFSRuntime
metadata:
  annotations:
    fluid.io/dataset-check-policy: "ignore"  # Core control flag, 
spec:

Why is this feature needed:
For some practical cases, users need to create Runtime without pre-defining a Dataset resource.

[cheyang]

What feature you'd like to add:

Background

In some corner cases, users need to create Runtime without pre-defining a Dataset resource. Current Fluid architecture requires strict Dataset-Runtime binding, which brings unnecessary complexity for temporary data scenarios. we propose allowing creation of runtime without Dataset dependency through Kubernetes annotations.

Annotation Design

apiVersion: data.fluid.io/v1alpha1
kind: JuiceFSRuntime
metadata:
  annotations:
    fluid.io/dataset-check-policy: "ignore"  # Core control flag, 
spec:

Why is this feature needed: For some practical cases, users need to create Runtime without pre-defining a Dataset resource.

Thank you for raising this proposal! 😊 Instead of bypassing Dataset checks entirely, we propose an auto-create policy to automatically generate minimal but valid Datasets. This maintains compatibility with Fluid’s core features while streamlining your workflow:

apiVersion: data.fluid.io/v1alpha1
kind: Runtime
metadata:
  annotations:
    fluid.io/dataset-policy: "auto-create"  # Auto-generates Dataset
spec:
  

Why auto-create :

1. Future-Proof Compatibility： Auto-created Datasets work seamlessly with DataLoad (preheating), DataMigrate, and Fluid’s monitoring stack.
2. Production-Ready: This model aligns with Fluid’s architecture, avoiding technical debt from diverging workflows.