radeonsi: stop using the VM_ALWAYS_VALID flag

Allocation all the bo as ALWAYS_VALID means they must all fit in memory
(vram + gtt) at each command submission.
This causes some trouble when the total allocated memory is greater than
the available memory.

Possible solutions:
- being able to tag/untag a bo as ALWAYS_VALID: would require kernel changes
- disable VM_ALWAYS_VALID when memory usage is more than a percentage of the
  available memory
- disable VM_ALWAYS_VALID entirely

v1 of this patch implemented option 2. v2 (this version) implements option 3.

Related issues:
 - https://gitlab.freedesktop.org/drm/amd/issues/607
 - https://gitlab.freedesktop.org/mesa/mesa/issues/1257

It also helps with some piglit tests (-t maxsize -t "max[_-].*size" -t maxuniformblocksize):
instead of crashing the machine, the tests fail cleanly.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2190
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3430>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/merge_requests/3430>

src/gallium/winsys/amdgpu/drm/amdgpu_cs.c

@@ -1398,9 +1398,6 @@ void amdgpu_cs_submit_ib(void *job, int thread_index)
 
       simple_mtx_lock(&ws->global_bo_list_lock);
       LIST_FOR_EACH_ENTRY(bo, &ws->global_bo_list, u.real.global_list_item) {
-         if (bo->is_local)
-            continue;
-
          list[num_handles].bo_handle = bo->u.real.kms_handle;
          list[num_handles].bo_priority = 0;
          ++num_handles;
@@ -1425,10 +1422,6 @@ void amdgpu_cs_submit_ib(void *job, int thread_index)
       unsigned num_handles = 0;
       for (i = 0; i < cs->num_real_buffers; ++i) {
          struct amdgpu_cs_buffer *buffer = &cs->real_buffers[i];
-
-         if (buffer->bo->is_local)
-            continue;
-
          assert(buffer->u.real.priority_usage != 0);
 
          list[num_handles].bo_handle = buffer->bo->u.real.kms_handle;