Ability to Autoremoval of orphaned Services or Nodes

User story:

• As a Developer, I'm trying and playing with my DB instances on local or DEV environments. I want them to be in PMM so I can track queries. I'm deleting the DB instances often, and they stay in my PMM. This making PMM overloaded with orphans and complicates my work. To clean Up PMM, I need manually delete them and agents related to them. This action takes time.
  I want to mark instances or nodes when I'm creating them that I'm ok with nodes/services being removed from the PMM if they are dead for some time.
• I'm a DBA deploying DBs in a robust environment, and I create a lot of DB instances but don't control the process of their removal. Some other 3rd party cares about this. Also DB instances can die at any moment (p.e. "spot instances")... it's ok, and our system is ready for this.
  But, as a result, I'm having PMM with a lot of dead DB instances. I need a way to make my PMM clean up the dead instances from the inventory.

Note: adding "autoremoval/unregister " for the Node itself is adding more value than for services

UI/UX:

*TBD *: Name of the flag :
autounregister-timeout, ?

#pmm-admin configure --autounregister-timeout ....
#pmm-admin add --autounregister-timeout ....

API and UI fields to specify autodestroy-timeout when a Service/Node is added.

Acceptance criteria

• PMM User able to specify flag `--autodestroy-timeout` (maybe ? "autounregister-timeout" ? ) in Time units on `pmm-admin configure` or `pmm-admin add` command
• PMM server removes Services or Nodes from the Inventory if the marked with flag and instances were down/not reporting data/ not connected (TBD)
• PMM user is able to specify timeout over API when registering/adding node and service to PMM
• PMM user is able to specify timeout for remote instances when adding them over UI

Out of scope:

• Making this a global setting for the PMM server
• Introducing global default timeout

Suggested implementation:
TBD

How to test:
TBD

Details:
TBD