Adapt DIRAC consumers to the binary status model(?)

[aldbr]

Description

Adapt the ResourceStatus and SiteStatus singletons, together with their callers, to work with the simplified binary status served by diracx (see phase 1, #837).

Before starting, confirm that the richer "information layer" (Active, Probing, Degraded, …) used today is not actually relied upon by any consumer and can be safely collapsed into the binary allowed/banned model. Any place that genuinely needs more than binary information should be called out explicitly so we can decide whether to adapt the consumer or to extend the model.

Scope

• Singleton-level changes in ResourceStatus and SiteStatus so that their public API keeps working when the underlying data is binary.
• Adjustments in callers that currently display or filter by raw status strings (e.g. SiteStatus.getSites(siteState=...)).
• Keeping the singleton cache format compatible with the binary data.

Reference

• Phase 1 main issue: [RSS Phase 1] Read-only resource status via CacheableSource #836
• Phase 1 data models: RSS data models (SQLAlchemy + Pydantic) #837
• Phase 2 discussion: #790 comment

Acceptance criteria

• Confirmation (with examples) that the binary model is sufficient for all current consumers, or an explicit list of consumers that need more.
• ResourceStatus and SiteStatus singletons keep their public API while operating on binary data.
• Callers that filter/display raw status strings are updated.
• Existing DIRAC unit tests still pass; new tests cover the adapted behavior where relevant.
• No change to the legacy RPC service or its DB schema.