gpu: nvgpu: unit: add channel close tests

Add branch coverage for:
- nvgpu_channel_kill
- nvgpu_channel_close

Also, modified gk20a_free_channel as follows:
- use nvgpu_assert to check ch->g (so that it can be tested)
- make sure g is non-NULL before calling nvgpu_get_poll_timeout

Jira NVGPU-3480

Change-Id: Ie1fa231b022da47b9ef9022ae67a6b3d73c28a8b
Signed-off-by: Thomas Fleury <tfleury@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/2129724
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>

userspace/units/fifo/nvgpu-fifo.c

@@ -28,6 +28,7 @@
 
 #include <nvgpu/channel.h>
 #include <nvgpu/tsg.h>
+#include <nvgpu/mm.h>
 #include <nvgpu/fifo/userd.h>
 #include <nvgpu/gk20a.h>