fix example to clean up watchdog goroutine

[raidancampbell]

This fixes the example code for cleaning up long-running VMs. Previously, they were killed after the timeout: the goroutine tracking this timeout holds a reference to the VM, therefore keeping the entire VM in memory for the duration of the configured timeout regardless of whether it has already completed its execution.

This change adds a second layer of communication to the watchdog goroutine, so that it will terminate itself if the VM has exited within the timeout. This allows the VM to be garbage collected if no other references to the VM exist.

[stevenh]

Thanks for this, in case its easier this is the full block with formatting fixed:

    vm := otto.New()
    vm.Interrupt = make(chan func(), 1) // The buffer prevents blocking
    watchdogCleanup := make(chan struct{})
    defer close(watchdogCleanup)

    go func() {
        select {
        case <-time.After(2 * time.Second): // Stop after two seconds
            vm.Interrupt <- func() {
                panic(halt)
            }
        case <-watchdogCleanup:
        }
        close(vm.Interrupt)
    }()

[raidancampbell]

heh, thanks for making my life easy by turning those into suggested changes: all have been applied