[postmortem] add postmortem metadata for symbols

As discussed in nodejs/llnode#156, we need
postmortem metadata for Symbols to properly print Symbol property names
in postmortem debugging tools. Patch suggested by Ben Noordhuis
(nodejs/llnode#156 (comment)).

[email protected], [email protected]

Change-Id: Ied6d3c079e8b23a9c796bc632c37785ed7dbc118
Reviewed-on: https://chromium-review.googlesource.com/1205052
Reviewed-by: Benedikt Meurer <[email protected]>
Reviewed-by: Yang Guo <[email protected]>
Commit-Queue: Yang Guo <[email protected]>
Cr-Commit-Position: refs/heads/master@{#55632}

BUILD.gn

@@ -859,6 +859,8 @@ action("postmortem-metadata") {
     "src/objects/js-regexp-string-iterator.h",
     "src/objects/map.h",
     "src/objects/map-inl.h",
+    "src/objects/name.h",
+    "src/objects/name-inl.h",
     "src/objects/scope-info.h",
     "src/objects/script.h",
     "src/objects/script-inl.h",

tools/gen-postmortem-metadata.py

@@ -417,15 +417,10 @@ def load_objects_from_file(objfilename, checktypes):
         # way around.
         #
         for type in types:
-                #
-                # Symbols and Strings are implemented using the same classes.
-                #
-                usetype = re.sub('SYMBOL_', 'STRING_', type);
-
                 #
                 # REGEXP behaves like REG_EXP, as in JS_REGEXP_TYPE => JSRegExp.
                 #
-                usetype = re.sub('_REGEXP_', '_REG_EXP_', usetype);
+                usetype = re.sub('_REGEXP_', '_REG_EXP_', type);
 
                 #
                 # Remove the "_TYPE" suffix and then convert to camel case,