fix(heal): honor explicit runtime state in status

[cxymds]

Problem

The RustFS background-heal API now reports an explicit runtime state, but the CLI still derived healing only from legacy scan counters. During a running heal this could make repeated status calls alternate between healing: false and healing: true as counters changed.

Solution

• Treat explicit state=active as authoritative and report healing=true.
• Treat explicit disabled, uninitialized, and idle states as authoritative and report healing=false, even if stale legacy counters remain.
• Preserve the legacy counter heuristic only when the server omits or returns an unknown state.
• Add regression tests for active state without legacy counters and disabled state with stale counters.

Verification

• cargo fmt --all --check
• cargo clippy --workspace -- -D warnings
• cargo test --workspace