PP: Non-functional: Only use string <-> atom mapping when needed.

Also, eliminate the 'atom' field of TPpToken.

Parsing a real 300 line shader, through to making the AST, is about 10% faster.

Memory is slightly reduced (< 1%).

The whole google-test suite, inclusive of all testing overhead, SPIR-V generation,
etc., runs 3% faster.

Since this is a code *simplification* that leads to perf. improvement, I'm not
going to invest too much more in measuring the perf. than this. The PP code is
simply now in a better state to see how to further rationalize/improve it.

glslang/MachineIndependent/preprocessor/PpAtom.cpp

@@ -120,7 +120,6 @@ const struct {
     { PpAtomIncrement,      "++" },
 
     { PpAtomDefine,         "define" },
-    { PpAtomDefined,        "defined" },
     { PpAtomUndef,          "undef" },
     { PpAtomIf,             "if" },
     { PpAtomElif,           "elif" },
@@ -149,17 +148,29 @@ const struct {
 
 namespace glslang {
 
+//
+// Map an existing string to an atom.
+//
+// Return 0 if no existing string.
+//
+int TPpContext::LookUpString(const char* s)
+{
+    auto it = atomMap.find(s);
+    return it == atomMap.end() ? 0 : it->second;
+}
+
 //
 // Map a new or existing string to an atom, inventing a new atom if necessary.
 //
 int TPpContext::LookUpAddString(const char* s)
 {
-    auto it = atomMap.find(s);
-    if (it == atomMap.end()) {
-        AddAtomFixed(s, nextAtom);
-        return nextAtom++;
-    } else
-        return it->second;
+    int atom = LookUpString(s);
+    if (atom == 0) {
+        atom = nextAtom++;
+        AddAtomFixed(s, atom);
+    }
+
+    return atom;
 }
 
 //