Cache the fallback font dictionary on the PartialEvaluator (PR 11218 follow-up)

This way we'll benefit from the existing font caching, and can thus avoid re-creating a fallback font over and over again during parsing.
(Thece changes necessitated the previous patch, since otherwise breakage could occur e.g. with fake workers.)

src/core/evaluator.js

@@ -935,11 +935,7 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
 
         // Falling back to a default font to avoid completely broken rendering,
         // but note that there're no guarantees that things will look "correct".
-        fontRef = new Dict();
+        fontRef = PartialEvaluator.getFallbackFontDict();
-        fontRef.set("BaseFont", Name.get("PDFJS-FallbackFont"));
-        fontRef.set("Type", Name.get("FallbackType"));
-        fontRef.set("Subtype", Name.get("FallbackType"));
-        fontRef.set("Encoding", Name.get("WinAnsiEncoding"));
       }
 
       if (this.fontCache.has(fontRef)) {
@@ -3134,6 +3130,21 @@ var PartialEvaluator = (function PartialEvaluatorClosure() {
     }
   };
 
+  // TODO: Change this to a `static` getter, using shadowing, once
+  //       `PartialEvaluator` is converted to a proper class.
+  PartialEvaluator.getFallbackFontDict = function() {
+    if (this._fallbackFontDict) {
+      return this._fallbackFontDict;
+    }
+    const dict = new Dict();
+    dict.set("BaseFont", Name.get("PDFJS-FallbackFont"));
+    dict.set("Type", Name.get("FallbackType"));
+    dict.set("Subtype", Name.get("FallbackType"));
+    dict.set("Encoding", Name.get("WinAnsiEncoding"));
+
+    return (this._fallbackFontDict = dict);
+  };
+
   return PartialEvaluator;
 })();