ci/tu: Skip more subgroup.clustered vector tests pre-merge.

These have been intermittently timing out on HDK8350/A660, showing up as
flakes (then the shader cache gets the re-run to pass, since there are
multiple shaders compiled per test).  Extending the pre-merge skip set on
the other devices should also help tame our runtimes, giving us more room
for other important testing (vkd3d, KHR_display).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/39568>

src/freedreno/ci/freedreno-a618-slow-skips.txt

@@ -63,4 +63,5 @@ spec@glsl-4.20@execution@conversion@frag-conversion-implicit-mat4-dmat4-zero-sig
 # Slow tests that may hit 60s timeout pre-merge, but pass otherwise.  94% CPU
 # used in ir3 create_variant, but no single pass there dominating its time.
 # Just use the scalar/vec2 cases, which are fast enough for CI.
-dEQP-VK.subgroups.clustered.compute.subgroupclustered(min|max|mul)_[iu]64vec[34].*
+dEQP-VK.subgroups.clustered.compute.subgroupclustered(min|max|mul|and|or|xor|add)_[iu]64vec[34].*
+dEQP-VK.subgroups.clustered.compute.subgroupclustered(min|max|mul|and|or|xor|add)_[iu]vec[34].*

src/freedreno/ci/freedreno-a660-slow-skips.txt

@@ -12,7 +12,7 @@ dEQP-VK.subgroups.clustered.compute.subgroupclusteredmin_uint64_t_requiredsubgro
 # Slow tests that may hit 60s timeout pre-merge, but pass otherwise.  94% CPU
 # used in ir3 create_variant, but no single pass there dominating its time.
 # Just use the scalar cases, which are fast enough for CI.
-dEQP-VK.subgroups.clustered.compute.subgroupclustered(min|max|mul)_[iu]64vec[234].*
+dEQP-VK.subgroups.clustered.compute.subgroupclustered(min|max|mul|and|or|xor|add)_[iu]64vec[234].*
 
 # Speed up pre-merge testing by removing redundant coverage in force-gmem mode
 gmem-dEQP-VK.api.*

src/freedreno/ci/freedreno-a750-slow-skips.txt

@@ -1,3 +1,8 @@
 # Note: skips lists for CI are just a list of lines that, when
 # non-zero-length and not starting with '#', will regex match to
 # delete lines from the test list.  Be careful.
+
+# Slow tests that aren't very useful coverage, but pass otherwise.  94% CPU
+# used in ir3 create_variant, but no single pass there dominating its time.
+# Just use the scalar/vec2 cases, which are fast enough for CI.
+dEQP-VK.subgroups.clustered.*.subgroupclustered(min|max|mul|and|or|xor|add)_[iu]64vec[34].*