[flang] Remove unneeded RoundDefault enumerator, and fix spelling in comments

flang/include/flang/Decimal/decimal.h

 };
 
 enum FortranRounding {
-  RoundNearest, /* RN */
+  RoundNearest, /* RN and RP */
   RoundUp, /* RU */
   RoundDown, /* RD */
   RoundToZero, /* RZ - no rounding */
   RoundCompatible, /* RC: like RN, but ties go away from 0 */
-  RoundDefault, /* RP: maps to one of the above */
 };
 
 /* The "minimize" flag causes the fewest number of output digits

flang/lib/Decimal/big-radix-floating-point.h

 
 public:
   explicit BigRadixFloatingPointNumber(
-      enum FortranRounding rounding = RoundDefault)
+      enum FortranRounding rounding = RoundNearest)
       : rounding_{rounding} {}
 
   // Converts a binary floating point value.
   explicit BigRadixFloatingPointNumber(
-      Real, enum FortranRounding = RoundDefault);
+      Real, enum FortranRounding = RoundNearest);
 
   BigRadixFloatingPointNumber &SetToZero() {
     isNegative_ = false;

   int digitLimit_{maxDigits}; // precision clamp
   int exponent_{0}; // signed power of ten
   bool isNegative_{false};
-  enum FortranRounding rounding_ { RoundDefault };
+  enum FortranRounding rounding_ { RoundNearest };
 };
 } // namespace Fortran::decimal
 #endif

flang/lib/Decimal/binary-to-decimal.cpp

     bool incr{false};
     switch (rounding_) {
     case RoundNearest:
-    case RoundDefault:
       incr = *end > '5' ||
           (*end == '5' && (p > end + 1 || ((end[-1] - '0') & 1) != 0));
       break;

flang/lib/Decimal/decimal-to-binary.cpp

   bool incr{false};
   switch (rounding_) {
   case RoundNearest:
-  case RoundDefault:
     incr = LSD > radix / 2 || (LSD == radix / 2 && digit_[0] % 2 != 0);
     break;
   case RoundUp:

   bool incr{false};
   switch (rounding) {
   case RoundNearest:
-  case RoundDefault:
     incr = guard > oneHalf || (guard == oneHalf && (fraction & 1));
     break;
   case RoundUp:

flang/lib/Evaluate/host.h

 // It should be defined when gcc/clang have a better support for it.
 
 template <> struct HostTypeHelper<Type<TypeCategory::Real, 4>> {
-  // IEE 754 64bits
+  // IEEE 754 64bits
   using Type = std::conditional_t<sizeof(float) == 4 &&
           std::numeric_limits<float>::is_iec559,
       float, UnsupportedType>;
 };
 
 template <> struct HostTypeHelper<Type<TypeCategory::Real, 8>> {
-  // IEE 754 64bits
+  // IEEE 754 64bits
   using Type = std::conditional_t<sizeof(double) == 8 &&
           std::numeric_limits<double>::is_iec559,
       double, UnsupportedType>;

 };
 
 template <> struct HostTypeHelper<Type<TypeCategory::Real, 16>> {
-  // IEE 754 128bits
+  // IEEE 754 128bits
   using Type = std::conditional_t<sizeof(long double) == 16 &&
           std::numeric_limits<long double>::digits == 113 &&
           std::numeric_limits<long double>::max_exponent == 16384,