Define configuration details for GPU enabled SaaS Linux runners

We need to decide on some configuration details for the GPU enabled SaaS Linux runners.

1. What GPU type to use.

   We're operating our runners in us-east1 region in the us-east1-c and us-east1-d zones. Unfortunately, some of the GPU types are not available in both these zones.

   GPU Type	us-east1-c	us-east1-d
   NVIDIA A100 80GB	❌	❌
   NVIDIA A100 40GB	❌	❌
   NVIDIA T4	🆗	🆗
   NVIDIA V100	🆗	❌
   NVIDIA Tesla P4	❌	❌
   NVIDIA Tesla P100	🆗	❌
   NVIDIA Tesla K80	🆗	🆗

2. How many GPUs we want to assign to each ephemeral VM that executes the job.