Stop using spot instances for release builds (#1760)

We now had a failure to build release binaries due to a spot instance
getting canceled early for the second time. For now, I propose not using
a spot instance. While this will increase the cost, with it only getting
run once per release, I don't think there's cause for concern.

This change is change is already deployed to the AWS account to build
the 1.4 release artifacts.

infra/spot_lambdas/start.py

@@ -22,13 +22,6 @@ def lambda_handler(event, context):
             "LaunchTemplateId": launch_template_id,
             "Version": "$Latest"
         },
-        InstanceMarketOptions={
-            "MarketType": "spot",
-            "SpotOptions": {
-                "SpotInstanceType": "one-time",
-                "InstanceInterruptionBehavior": "terminate",
-            },
-        },
         TagSpecifications=[
             {
                 "ResourceType": "instance",
@@ -49,4 +42,4 @@ def lambda_handler(event, context):
         'body': {
             "instance_id": instance_id,
         }
-    }
+    }