Pre-cache PLY lex table and use optimized mode (5% build time improvement)

Source/bindings/generated_bindings.gyp

 #
 # Copyright (C) 2013 Google Inc. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions are
 # met:
 #
 #     * Redistributions of source code must retain the above copyright
 # notice, this list of conditions and the following disclaimer.
 #     * Redistributions in binary form must reproduce the above
@@ skipping 266 matching lines @@
         '<(SHARED_INTERMEDIATE_DIR)/blink/InterfacesInfo.pickle',
         '--event-names-file',
         '<(SHARED_INTERMEDIATE_DIR)/blink/EventInterfaces.in',
         '<@(write_file_only_if_changed)',
       ],
       'message': 'Generating list of Event interfaces',
       }]
   },
 ################################################################################
   {
-    # A separate pre-caching step is *not required* to use parse table caching
-    # in PLY, as the cache is concurrency-safe.
+    # A separate pre-caching step is *not required* to use lex/parse table
+    # caching in PLY, as the caches are concurrency-safe.
     # However, pre-caching ensures that all compiler processes use the cached
-    # file (hence maximizing speed), instead of early processes building the
-    # table themselves (as it's not yet been written by the time they start).
+    # files (hence maximizing speed), instead of early processes building the
+    # tables themselves (as they've not yet been written when they start).
     'target_name': 'cached_yacc_tables',
     'type': 'none',
     'actions': [{
       'action_name': 'cache_yacc_tables',
       'inputs': [
         '<@(idl_lexer_parser_files)',
       ],
       'outputs': [
+        '<(bindings_output_dir)/lextab.py',
         '<(bindings_output_dir)/parsetab.pickle',
       ],
       'action': [
         'python',
         'scripts/blink_idl_parser.py',
         '<(bindings_output_dir)',
       ],
-      'message': 'Caching PLY yacc parse tables',
+      'message': 'Caching PLY lex & yacc lex/parse tables',
     }],
   },
 ################################################################################
   {
     # A separate pre-caching step is *required* to use bytecode caching in
     # Jinja (which improves speed significantly), as the bytecode cache is
     # not concurrency-safe on write; details in code_generator_v8.py.
     'target_name': 'cached_jinja_templates',
     'type': 'none',
     'actions': [{
@@ skipping 31 matching lines @@
     'sources': [
       '<@(interface_idl_files)',
     ],
     'rules': [{
       'rule_name': 'binding',
       'extension': 'idl',
       'msvs_external_rule': 1,
       'inputs': [
         '<@(idl_lexer_parser_files)',  # to be explicit (covered by parsetab)
         '<@(idl_compiler_files)',
+        '<(bindings_output_dir)/lextab.py',
         '<(bindings_output_dir)/parsetab.pickle',
         '<(bindings_output_dir)/cached_jinja_templates.stamp',
         'IDLExtendedAttributes.txt',
         # If the dependency structure or public interface info (e.g.,
         # [ImplementedAs]) changes, we rebuild all files, since we're not
         # computing dependencies file-by-file in the build.
         # This data is generally stable.
         '<(SHARED_INTERMEDIATE_DIR)/blink/InterfacesInfo.pickle',
         # Further, if any dependency (partial interface or implemented
         # interface) changes, rebuild everything, since every IDL potentially
@@ skipping 56 matching lines @@
     'type': 'none',
     'dependencies': [
       'aggregate_generated_bindings',
       'event_interfaces',
       'individual_generated_bindings',
     ],
   },
 ################################################################################
   ],  # targets
 }