Fix flaky runWithRateLimit test by using fake timers

[gonzaloriestra]

WHY are these changes introduced?

Flaky test: https://github.com/Shopify/cli/actions/runs/24396106090

FAIL   @shopify/cli-kit  src/private/node/conf-store.test.ts > runWithRateLimit > throttles the task when the cache is populated recently
AssertionError: expected true to be false // Object.is equality

- Expected
+ Received

- false
+ true

 ❯ src/private/node/conf-store.test.ts:506:19
    504| 
    505|       // Then
    506|       expect(got).toBe(false)
       |                   ^
    507|       expect(taskRan).toBe(false)
    508|     })
 ❯ Module.inTemporaryDirectory src/public/node/fs.ts:81:12
 ❯ src/private/node/conf-store.test.ts:475:5

The 'throttles the task when the cache is populated recently' test was flaky because it used real time with a 1-second timeout window. Under CI load, the time between filling the rate limit cache and checking it could exceed 1 second, causing the old occurrences to be swept out and the rate limit to no longer be exceeded.

WHAT is this pull request doing?

Fix by using vi.useFakeTimers() to freeze time during the test, matching the pattern already used by the 'cache is populated but outdated' test in the same describe block. The existing afterEach hook already calls vi.useRealTimers() for cleanup.

Also applied the same fix to the 'rate limit isn't used up' test to prevent the same class of flake.

How to test your changes?

CI

Checklist

• I've considered possible cross-platform impacts (Mac, Linux, Windows)
• I've considered possible documentation changes
• I've considered analytics changes to measure impact
• The change is user-facing, so I've added a changelog entry with pnpm changeset add