SCMU does not set ServiceControlQueueAddress when only adding audit instances bug fix

[cquirosj]

Feature: Audit Instance ServiceControl Queue Address

As an operator adding instances through ServiceControl Management Utility (SCMU), I want every audit instance to be connected to a ServiceControl (error) instance so that messages the audit instance sends to the error instance always have a valid destination.

Bug: #4753 — SCMU does not set ServiceControlQueueAddress when only adding audit instances.
When the ServiceControl.Audit/ServiceControlQueueAddress setting is missing from
ServiceControl.Audit.exe.config, the audit instance fails at runtime with
"no destination specified for message".

Rules and Examples

Rule 1: Must address the audit instance to the error instance installed in the same session

When the user installs an error instance and an audit instance together, the audit
instance's queue address is the name of the error instance being installed — never an
already-installed one.

• Example: The one where both instances are installed together and the audit
  instance's queue address is the new error instance's name.
• Counter-example: The one where other error instances already exist on the
  machine, yet no choice is offered — the error instance being installed always wins.

---

Rule 2: Should auto-detect the existing error instance when adding an audit instance alone

• Example: The one where exactly one error instance exists on the machine and its
  name is used as the queue address without any user input (no dropdown shown).

---

Rule 3: Must require an explicit choice when multiple existing error instances are found

Auto-detection cannot guess between several error instances — picking one silently
risks routing messages to the wrong instance. The choice dropdown is shown only in
this case.

• Example: The one where two error instances exist and the dropdown offers both.
• Example: The one where Save is blocked until the user picks one of the detected
  instances, and unblocked once a choice is made.

---

Rule 4: Must require a manually entered queue address when no error instance is detected

Detection only sees instances installed on the local machine, but the error instance
may legitimately live elsewhere — e.g. several VMs each hosting an audit instance that
feeds one central error instance for load balancing. When nothing is detected locally,
the user must supply the queue address of the (possibly remote) error instance; Save
stays blocked until a value is entered so a config entry is always written.

• Example: The one where no error instance is detected and Save is blocked until
  the user enters the remote error instance's address, then unblocked once entered
  and the entered value is used as the queue address.
• Example: The one where the entered address contains whitespace and is rejected
  (instance names — and therefore their queue addresses — never contain whitespace).
• Counter-example: The one where only an error instance is being installed — the
  queue address does not apply, no entry field is shown, and no validation error is
  raised.