Implement @harrishancock's horrible work-around for anyOfChars() ICE on MSVC.

I think this is worthwhile since otherwise this header can't be used at all on MSVC, which has all sorts of sad downstream effects. Hopefully we can revert this soon, when MSVC gets fixed.

c++/src/kj/parse/char.h

@@ -152,6 +152,13 @@ constexpr inline CharGroup_ charRange(char first, char last) {
   return CharGroup_().orRange(first, last);
 }
 
+#if _MSC_VER
+#define anyOfChars(chars) CharGroup_().orAny(chars)
+// TODO(msvc): MSVC ICEs on the proper definition of `anyOfChars()`, which in turn prevents us from
+//   building the compiler or schema parser. We don't know why this happens, but Harris found that
+//   this horrible, horrible hack makes things work. This is awful, but it's better than nothing.
+//   Hopefully, MSVC will get fixed soon and we'll be able to remove this.
+#else
 constexpr inline CharGroup_ anyOfChars(const char* chars) {
   // Returns a parser that accepts any of the characters in the given string (which should usually
   // be a literal).  The returned parser is of the same type as returned by `charRange()` -- see
@@ -159,6 +166,7 @@ constexpr inline CharGroup_ anyOfChars(const char* chars) {
 
   return CharGroup_().orAny(chars);
 }
+#endif
 
 // =======================================================================================