Stabilize dashboard token provenance tests

[Yuxin-Qiao]

Summary

• inject each test fixture UserDefaults into the Dashboard controller
• keep Token Provenance tests isolated from the process default settings
• complement test(repo-hygiene): restore macOS test baseline #2418, which applies the same fix to SpendDashboardControllerTests

Why

The Dashboard controller otherwise reads process defaults while its request builder reads the fixture settings. That mismatch causes the test to observe empty groups despite a confirmed fixture snapshot.

Validation

• SpendDashboardTokenProvenanceTests: 13 tests passed
• make check: passed
• full make test: not rerun on this standalone branch because it still lacks the separate Controller fixes in test(repo-hygiene): restore macOS test baseline #2418. Together these PRs cover all 13 group-49 failures reproduced from main.

UI proof

Not applicable: test-isolation change only.

Provider / privacy

No production provider behavior, credentials, or account data access changed.

[clawsweeper]

ClawSweeper status: review started.

I am starting a fresh review of this pull request: Stabilize dashboard token provenance tests This is item 1/1 in the current shard. Shard 0/1.

This placeholder means the worker is alive and reading the current context. I will edit this same comment with the actual review when the claws are done clicking.

Crustacean status: shell secured, claws on keyboard, evidence pebbles being sorted.

[Yuxin-Qiao]

Superseded by #2418, which now contains this Token Provenance test-isolation fix together with the complete macOS test-baseline restoration and full validation.