cmdlib: workaround rofiles-fuse mounts leaking

This started happening in CI:

```
+ mount -o remount,ro /srv/cache
mount: /srv/cache: mount point is busy.
       dmesg(1) may have more information after failed mount system call.
[  321.026584] Kernel panic - not syncing: Attempted to kill init! exitcode=0x00002000
```

Digging into it, it seems to be due to leftover rofiles-fuse mounts:

```
+ ps aux
...
root       306  0.0  0.1 401092  3972 ?        Ssl  15:46   0:00 rofiles-fuse --copyup usr /tmp/rpmostree-rofiles-fuse0QroLi
root       311  0.0  0.1 251564  2488 ?        Ssl  15:46   0:00 rofiles-fuse --copyup etc /tmp/rpmostree-rofiles-fuseANAAgx
root       330  0.0  0.2 474828  4120 ?        Ssl  15:46   0:00 rofiles-fuse --copyup usr /tmp/rpmostree-rofiles-fuseioSZ7r
root       334  0.0  0.1 251564  2712 ?        Ssl  15:46   0:00 rofiles-fuse --copyup etc /tmp/rpmostree-rofiles-fusePPGBR9
...
```

For some reason, rpm-ostree is not able to unmount those.

We should dig into this but as a brutal workaround, just kill them
manually. There's no data loss concerns here; the commit has already
been written and shipped out via virtiofs.

The issue remains tracked at
#3848.

src/cmdlib.sh

@@ -758,6 +758,8 @@ else
 fi
 echo \$rc > ${rc_file}
 if [ -n "\${cachedev}" ]; then
+    # XXX: brutal workaround for https://github.com/coreos/coreos-assembler/issues/3848
+    killall rofiles-fuse || :
     /sbin/fstrim -v ${workdir}/cache
     mount -o remount,ro ${workdir}/cache
     fsfreeze -f ${workdir}/cache