ci/turnip: Extend a630 vk full timeout to 3 hours.

With the CTS coverage and tu featureset extending, these jobs have been
reliably timing out for a while.  I've updated the xfails based on a
single run, we'll see how that goes in the upcoming nightlies.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21879>

src/freedreno/ci/gitlab-ci.yml

@@ -297,10 +297,11 @@ a630_vk:
     DEQP_SUITE: freedreno-a630-vk
 
 a630_vk_full:
-  # We use a longer timeout (2 hour job, 110 minute deqp) to keep the parallel
-  # down so that we don't lock up too many runners for a long time when a dev is
-  # testing full VK status.
-  timeout: 2h
+  # We use a longer timeout (3 hour job) to keep the parallel down so that we
+  # don't lock up too many runners for a long time when a dev is testing full VK
+  # status.  The full runs are restricted to just 2 runners to keep from
+  # blocking up normal merges, so going more parallel doesn't make any sense.
+  timeout: 3h
   extends:
     - a630_vk
     - .a630-full
@@ -308,7 +309,7 @@ a630_vk_full:
   parallel: 2
   variables:
     DEQP_SUITE: freedreno-a630-vk-full
-    TEST_PHASE_TIMEOUT: 110
+    TEST_PHASE_TIMEOUT: 180
 
 a630_vk_asan:
   extends: