Removing ephemeral experiment from scale-config.yml and updating validate_scale_config.py (#6605)

With the success of ephemeral experimentation, running in all branches,
nightly and trunk. We maintained a long cool off period and [executed
multiple failover
scenarios](https://docs.google.com/document/d/1ttAsjMrCEoEyqnIs5UdxzkAvL7xKnO10ytsm7hg9rWQ/edit?tab=t.4fwxjpm6qaus#heading=h.1y7utpose46a)
to make sure the system is stable even when there is catastrophic
failure in any dependent system.

This lead us to safely stop the `ephemeral` experiment in the issue
#5132 and then remove all
ephemeral variants for the runners.

Next step would be renaming all runners that have `.ephemeral` suffix,
as it is redundant at this stage.

Author: jeanschmidt

.github/canary-scale-config.yml

@@ -37,309 +37,198 @@ runner_types:
     instance_type: m7i-flex.8xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.12xlarge:
     disk_size: 200
     instance_type: c5.12xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.10xlarge.avx2:
     disk_size: 200
     instance_type: m4.10xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.24xl.spr-metal:
     disk_size: 200
     instance_type: c7i.metal-24xl
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.16xlarge.spr:
     disk_size: 200
     instance_type: c7i.16xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.9xlarge.ephemeral:
     disk_size: 200
     instance_type: c5.9xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral: {}
   c.linux.12xlarge.ephemeral:
     disk_size: 200
     instance_type: c5.12xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral: {}
   c.linux.16xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.16xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.24xlarge:
     disk_size: 150
     instance_type: c5.24xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.24xlarge.ephemeral:
     disk_size: 150
     instance_type: c5.24xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral: {}
   c.linux.2xlarge:
     disk_size: 150
     instance_type: c5.2xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.4xlarge:
     disk_size: 150
     instance_type: c5.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.4xlarge.for.testing.donotuse:
     disk_size: 150
     instance_type: c5.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.4xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.8xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.8xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g4dn.12xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g4dn.12xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g4dn.4xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g4dn.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g4dn.metal.nvidia.gpu:
     disk_size: 150
     instance_type: g4dn.metal
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g5.48xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g5.48xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g5.12xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g5.12xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g5.4xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g5.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.g6.4xlarge.experimental.nvidia.gpu:
     disk_size: 150
     instance_type: g6.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.large:
     disk_size: 15
     instance_type: c5.large
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.arm64.2xlarge:
     disk_size: 256
     instance_type: t4g.2xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.arm64.m7g.4xlarge:
     disk_size: 256
     instance_type: m7g.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.arm64.2xlarge.ephemeral:
     disk_size: 256
     instance_type: t4g.2xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral: {}
   c.linux.arm64.m7g.4xlarge.ephemeral:
     disk_size: 256
     instance_type: m7g.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral: {}
   c.linux.arm64.m7g.metal:
     disk_size: 256
     instance_type: m7g.metal
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.windows.g4dn.xlarge:
     disk_size: 256
     instance_type: g4dn.xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral: {}
   c.windows.g4dn.xlarge.nonephemeral:
     disk_size: 256
     instance_type: g4dn.xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.windows.4xlarge:
     disk_size: 256
     instance_type: c5d.4xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral: {}
   c.windows.4xlarge.nonephemeral:
     disk_size: 256
     instance_type: c5d.4xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.windows.8xlarge.nvidia.gpu:
     disk_size: 256
     instance_type: p3.2xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral: {}
   c.windows.8xlarge.nvidia.gpu.nonephemeral:
     disk_size: 256
     instance_type: p3.2xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.windows.g5.4xlarge.nvidia.gpu:
     disk_size: 256
     instance_type: g5.4xlarge
     is_ephemeral: true
     os: windows
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.2xlarge.memory:
     disk_size: 200
     instance_type: r5.2xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.4xlarge.memory:
     disk_size: 300
     instance_type: r5.4xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.8xlarge.memory:
     disk_size: 400
     instance_type: r5.8xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.12xlarge.memory:
     disk_size: 600
     instance_type: r5.12xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral:
-        is_ephemeral: true
   c.linux.12xlarge.memory.ephemeral:
     disk_size: 400
     instance_type: r5.12xlarge
     is_ephemeral: true
     os: linux
-    variants:
-      ephemeral: {}