Revert "[perfcompare] Add new entry point for perfcompare recipe to invoke"

This reverts commit f26b0a7a35cced57c0ce7eab63df5c1f53221027.

Reason for revert: appears to break vg bots in gi

Original change's description:
> [perfcompare] Add new entry point for perfcompare recipe to invoke
> 
> Add benchmarks_perfcompare.sh.
> 
> This initial version runs a subset of the tests that benchmarks.sh
> runs, because the full set takes too much time for the bot to run
> twice without going over the bot's current timeout limit.  Running a
> faster set of tests is also easier for development testing of the
> perfcompare recipe.
> 
> Bug: IN-646
> Test: tested via 'led' with recipe change
> Change-Id: I54349d462423cd89dc051f0ae69d91f908793842

TBR=mseaborn@google.com,nathanrogers@google.com

Change-Id: I890aed3617eb9b811985a7841390549f612f8018
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: IN-646
diff --git a/garnet/tests/benchmarks/BUILD.gn b/garnet/tests/benchmarks/BUILD.gn
index a307a2b..6b1fa41 100644
--- a/garnet/tests/benchmarks/BUILD.gn
+++ b/garnet/tests/benchmarks/BUILD.gn
@@ -16,9 +16,5 @@
       name = rebase_path("gfx_benchmarks.sh")
       dest = "gfx_benchmarks.sh"
     },
-    {
-      name = rebase_path("benchmarks_perfcompare.sh")
-      dest = "benchmarks_perfcompare.sh"
-    },
   ]
 }
diff --git a/garnet/tests/benchmarks/benchmarks_perfcompare.sh b/garnet/tests/benchmarks/benchmarks_perfcompare.sh
deleted file mode 100644
index 0a868d0..0000000
--- a/garnet/tests/benchmarks/benchmarks_perfcompare.sh
+++ /dev/null
@@ -1,33 +0,0 @@
-#!/boot/bin/sh
-#
-# Copyright 2019 The Fuchsia Authors. All rights reserved.
-# Use of this source code is governed by a BSD-style license that can be
-# found in the LICENSE file.
-#
-# This script is used for the performance comparison (perfcompare) CQ
-# bots, which compare performance before and after a change.
-#
-# This script runs a subset of benchmarks for the Garnet layer.  It
-# runs a subset of what benchmarks.sh runs.  The reason for running a
-# subset is that the full set of tests currently takes too long and
-# tends to exceed the bot timeout.
-#
-# For usage, see runbench_read_arguments in runbenchmarks.sh.
-
-# Import the runbenchmarks library.
-. /pkgfs/packages/runbenchmarks/0/data/runbenchmarks.sh
-
-runbench_read_arguments "$@"
-
-# Performance tests implemented in the Zircon repo.
-runbench_exec "${OUT_DIR}/zircon.perf_test.json" \
-    /system/test/sys/perf-test -p --out="${OUT_DIR}/zircon.perf_test.json"
-
-# Performance tests implemented in the Garnet repo (the name
-# "zircon_benchmarks" is now misleading).
-runbench_exec "${OUT_DIR}/zircon_benchmarks.json" \
-    /pkgfs/packages/zircon_benchmarks/0/test/zircon_benchmarks \
-    -p --out="${OUT_DIR}/zircon_benchmarks.json"
-
-# Exit with a code indicating whether any errors occurred.
-runbench_finish "${OUT_DIR}"