client: Don't invoke systemctl start if unit is already active

Adding onto the pile of hacks here unfortunately.  Basically
RHEL8 systemd seems to count explicit `systemctl start` invocations
against the restart limit.  We hit this in tests in openshift/machine-config-operator
which are invoking `rpm-ostree status` and `rpm-ostree kargs` frequently.

rust/src/client.rs

@@ -156,12 +156,30 @@ pub(crate) fn client_handle_fd_argument(arg: &str, arch: &str) -> CxxResult<Vec<
 /// Basically we load too much data before claiming the bus name,
 /// and dbus doesn't give us a useful error.  Instead, let's talk
 /// to systemd directly and use its client tools to scrape errors.
+///
+/// What we really should do probably is use native socket activation.
 pub(crate) fn client_start_daemon() -> CxxResult<()> {
     let service = "rpm-ostreed.service";
     // Assume non-root can't use systemd right now.
     if rustix::process::getuid().as_raw() != 0 {
         return Ok(());
     }
+    // Unfortunately, RHEL8 systemd will count "systemctl start"
+    // invocations against the restart limit, so query the status
+    // first.
+    let activeres = Command::new("systemctl")
+        .args(&["is-active", "rpm-ostreed"])
+        .output()?;
+    if !activeres.status.success() {
+        return Err(format!(
+            "Failed to run systemctl is-active rpm-ostreed: {}",
+            activeres.status
+        )
+        .into());
+    }
+    if String::from_utf8_lossy(&activeres.stdout).starts_with("active") {
+        return Ok(());
+    }
     let res = Command::new("systemctl")
         .args(&["--no-ask-password", "start", service])
         .status()?;