[BUG] AbstractDataChangedListener cache-update methods scan ALL namespaces, not just the changed one

## Description
`updateSelectorCache` -> `selectorService.listAll()`; `updateRuleCache` -> `ruleService.listAll()`; `updateAppAuthCache` -> `appAuthService.listAll()`; `updateDiscoveryUpstreamDataCache` -> `discoveryUpstreamService.listAll()`; `updateProxySelectorDataCache` -> `proxySelectorService.listAll()`; `updateAiProxyApiKeyCache` -> `aiProxyApiKeyService.listAll()`. All call the no-arg `listAll()` (ALL namespaces), even though `updateMetaDataCache` correctly uses `listAll(namespaceId)`. So a single selector change in namespace A triggers a full-table scan across all namespaces.

## Location
```
shenyu-admin/.../listener/AbstractDataChangedListener.java:353-395
```

## Impact
Work scales with total system size, not the changed namespace. Wasted DB + CPU proportional to total config on every change.

## Suggested fix
Use `listAllByNamespaceId(...)` for all groups (the methods exist).

## Related existing issue(s)
None

_Identified during the 2026-08-02 audit; full list in [`docs/issue-candidates-2026-08-02.md`](docs/issue-candidates-2026-08-02.md)._

Start in shenyu-admin/.../listener/AbstractDataChangedListener.java:353-395 and compare the affected cache-update methods with updateMetaDataCache. Inspect the existing listAllByNamespaceId(...) methods for each service, then verify that a change in one namespace no longer scans other namespaces and that the affected cache updates still work.

Written by the indexing model from the issue text.