Run shell commands synchronously and asynchronously on a Runloop devbox, with streaming output and stateful shells

Documented steps

1. Synchronous: Python `result = await devbox.cmd.exec("echo Hello World")`; read `await result.stdout()` and `result.exit_code`. TypeScript: `devbox.cmd.exec("echo Hello World")` with `result.exitCode`.
2. Asynchronous/background: Python `command = await devbox.cmd.exec_async("HOST=0.0.0.0; npx run http-server")`, then `await command.kill()` to stop it. TypeScript: `devbox.cmd.execAsync(...)` / `command.kill()`.
3. Stream live output by passing stdout/stderr callbacks: Python `devbox.cmd.exec_async(cmd, stdout=lambda x: print(x), stderr=lambda x: print(x))`; TypeScript passes `{ stdout: (x)=>..., stderr: (x)=>... }` as an options object.
4. By default each exec() call runs in an isolated shell session — environment variables set in one call do NOT persist to the next call.
5. To persist state (env vars, cwd) across multiple commands, pass the same `shell_name` parameter, e.g. Python `await devbox.cmd.exec("export MY_VAR=123", shell_name="my-shell")` then `await devbox.cmd.exec("echo $MY_VAR", shell_name="my-shell")`.
6. Docs: https://docs.runloop.ai/docs/devboxes/execute-commands