[infra] Add and enable nci_as_highest_tier variant

With work on NCI proceeding, it makes sense to test multiple
pipeline configurations.

The nci variant (passes --turbo-nci) now spawns dedicated NCI
compilation jobs and inserts generated code into the code cache.

The nci_as_highest_tier variant (passes --turbo-nci-as-highest-tier)
simply replaces TF with NCI code (no extra jobs, no extra caching).
This mode stresses NCI generated code more than the nci variant, in
which NCI code only runs on cache hits.

Bug: v8:8888
Change-Id: I4c2a43cce5271a6c288e7aba195dcc9daed6af9d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2299361
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68964}

infra/testing/builders.pyl

@@ -330,6 +330,7 @@
       {'name': 'v8testing', 'variant': 'infra_staging', 'shards': 2},
       # Native context independent code.
       {'name': 'v8testing', 'variant': 'nci'},
+      {'name': 'v8testing', 'variant': 'nci_as_highest_tier'},
       # Stress sampling.
       {'name': 'mjsunit', 'variant': 'stress_sampling'},
       {'name': 'webkit', 'variant': 'stress_sampling'},
@@ -960,6 +961,7 @@
       {'name': 'v8testing', 'variant': 'infra_staging', 'shards': 2},
       # Native context independent code.
       {'name': 'v8testing', 'variant': 'nci'},
+      {'name': 'v8testing', 'variant': 'nci_as_highest_tier'},
       # Stress sampling.
       {'name': 'mjsunit', 'variant': 'stress_sampling'},
       {'name': 'webkit', 'variant': 'stress_sampling'},
@@ -989,6 +991,7 @@
       {'name': 'v8testing', 'variant': 'infra_staging'},
       # Native context independent code.
       {'name': 'v8testing', 'variant': 'nci'},
+      {'name': 'v8testing', 'variant': 'nci_as_highest_tier'},
       # Stress sampling.
       {'name': 'mjsunit', 'variant': 'stress_sampling'},
       {'name': 'webkit', 'variant': 'stress_sampling'},

test/cctest/cctest.status

@@ -636,7 +636,7 @@
 }],
 
 ################################################################################
-['variant == nci', {
+['variant == nci or variant == nci_as_highest_tier', {
   # Optimizes and deopts differently than TurboFan.
   'test-api/FastApiCalls': [SKIP],
   'test-cpu-profiler/Deopt*': [SKIP],
@@ -650,6 +650,6 @@
   'test-heap/OptimizedPretenuring*': [SKIP],
   'test-heap-profiler/SamplingHeapProfilerPretenuredInlineAllocations': [SKIP],
   'test-log/LogAll': [SKIP],
-}],  # variant == nci
+}],  # variant == nci or variant == nci_as_highest_tier
 
 ]

test/debugger/debugger.status

@@ -136,10 +136,10 @@
 }], # lite_mode or variant == jitless
 
 ##############################################################################
-['variant == turboprop or variant == nci', {
+['variant == turboprop or variant == nci or variant == nci_as_highest_tier', {
   # Deopts differently than TurboFan.
   'debug/debug-optimize': [SKIP],
-}],  # variant == turboprop
+}],  # variant == turboprop or variant == nci or variant == nci_as_highest_tier
 
 ##############################################################################
 # Liftoff needs to be enabled before running these tests.

test/mjsunit/mjsunit.status

@@ -1298,7 +1298,7 @@
 }],  # variant == slow_path
 
 ################################################################################
-['variant == nci', {
+['variant == nci or variant == nci_as_highest_tier', {
   # Deopts differently than TurboFan.
   'compiler/is-being-interpreted*': [SKIP],
   'compiler/number-comparison-truncations': [SKIP],
@@ -1423,5 +1423,6 @@
   'compiler/serializer-feedback-propagation-1': [SKIP],
   'compiler/serializer-feedback-propagation-2': [SKIP],
   'compiler/serializer-transition-propagation': [SKIP],
-}],  # variant == nci
+}],  # variant == nci or variant == nci_as_highest_tier
+
 ]

tools/testrunner/local/variants.py

@@ -15,6 +15,7 @@ ALL_VARIANT_FLAGS = {
   "jitless": [["--jitless"]],
   "minor_mc": [["--minor-mc"]],
   "nci": [["--turbo-nci"]],
+  "nci_as_highest_tier": [["--turbo-nci-as-highest-tier"]],
   "no_lfa": [["--no-lazy-feedback-allocation"]],
   # No optimization means disable all optimizations. OptimizeFunctionOnNextCall
   # would not force optimization too. It turns into a Nop. Please see