Missing bearer token on proxy assets

## Issue

When loading external (app service) source code with `@equinor/fusion-framework-vite-plugin-spa` the service-worker will add bearer token to the initial request, then rewrite the request to the local proxy route, which will be the `import.url` for the loaded script. If the loaded script has assets or code splitting, no token will apply since it skips the service-worker and executes request directly to `@equinor/fusion-framework-vite-plugin-api-service`.

## Reproduce

use the CLI and load an external application (example `ffc portal dev` and navigate to `/apps/some-app-with-assets`) and the resources loaded from the initial app script will be missing bearer token in headers

## Proposed Solution

### Option 1

Add a proxy-host token set to store and re-use tokens to that host if bearer is missing in request.

> [!warning]
> - all requests to `@equinor/fusion-framework-vite-plugin-api-service` to that proxy-host will include token, _might not what we want?__
> - if request is executed before snatched before service-worker re-write will not obtain token, __should not happen__

### Option 2

remove the service-worker and use `@equinor/fusion-framework-module-msal-node` to authenticate the user before starting the dev-server

> [!warning]
> the front-end app registration does not support node login and uses an own CLI application registration. This might lead to different access policies than "real world".