[Bug]: Instances shown as Failed in Verda are Provisioning in dstack #3746

Steps to reproduce

1. Provision many Verda instances so that some of them are failed due to no capacity, e.g.

type: fleet
name: test-verda-fleet
nodes:
  min: 0
  target: 30
backends: [verda]
resources:
   cpu: 1..
   memory: 1GB..
   gpu: 0
placement: any
spot_policy: on-demand

2. Even if some instance are failed in Verda they are still Provisioning in dstack which means the fleet and instances stay in Provisioning until the provisioning timeout is due (minutes=10)

Actual behaviour

Not confirmed but possible consequence of #3723

Expected behaviour

update_provisionig_data() fails the instance by raising ProvisioningError when the instance in failed or terminated in the provider.

dstack version

master

Server logs

Additional information

No response