To deploy a new service that automatically scales using a managed instance group across multiple regions and requires a large number of resources for each instance, the following steps are crucial:

1. Validate resource requirements against project quota limits (Option D): Before deployment, it's essential to ensure that the resource requirements do not exceed the available project quota limits in each region to avoid deployment failures.
2. Use an appropriate machine type (Option B): The n2-highcpu-96 machine type is suitable for services requiring a large number of resources, as it provides high CPU capacity. Monitoring Cloud Trace (Option A) is more about optimizing performance post-deployment rather than planning for capacity. Deploying in multiple regions and using an internal load balancer (Option C) is about ensuring high availability and distributing traffic but doesn't directly address the need for planning capacity or selecting the right machine type. Therefore, the correct actions are to validate project quota limits and select an appropriate machine type.