Test command with Maestro runs ~6s faster when Maestro Studio is open

**Describe the bug**
When running a test command with Maestro, the test runs around 6 seconds faster if Maestro Studio is open with the test device connected.

**To Reproduce**

**Command:**

`flashlight test --bundleId com.pizzaapp --testCommand "MAESTRO_APP_ID=com.pizzaapp npx @perf-profiler/maestro@latest test .maestro/flows/Login.yaml"`

**Test file:**

https://github.com/infinitered/ignite/blob/master/boilerplate/.maestro/flows/Login.yaml

Steps to reproduce the behavior:

1. `npx ignite-cli@latest new PizzaApp --yes`
2. `yarn android`
3. Open Maestro Studio and select the device that you'll be running the test in
4. `flashlight test --bundleId com.pizzaapp --testCommand "MAESTRO_APP_ID=com.pizzaapp npx @perf-profiler/maestro@latest test .maestro/flows/Login.yaml"`
5. Observe Average Test Runtime
6. Close Maestro Studio
7. `flashlight test --bundleId com.pizzaapp --testCommand "MAESTRO_APP_ID=com.pizzaapp npx @perf-profiler/maestro@latest test .maestro/flows/Login.yaml"`
8. Observe that Average Test Runtime is ~6s longer

**Screenshots**

**Environment (please complete the following information):**

- OS: macOS 26.0.1
- Flashlight Version: 0.18.0
- Android SDK platform tools version: Android Debug Bridge version 1.0.41 Version 36.0.0-13206524 Running on Darwin 25.0.0 (arm64)
- Test Framework: Maestro v2.0.3
- Share the bundle identifier of your app: com.pizzaapp

**Additional context**
This might be more of a Maestro issue. I just wanted to share it here just in case others were having similar issues with test runtime inconsistencies.

Reproduce the command with Maestro Studio open and closed, using the .maestro/flows/Login.yaml test referenced in the report, and compare the average runtimes. Determine whether the timing difference originates in Flashlight or Maestro; done means documenting a reproducible cause or confirming that the behavior belongs to Maestro.

Written by the indexing model from the issue text.