Use char.IsAsciiLetter in the Regex runtime (dotnet#77559)

* Use char.IsAsciiLetter in the Regex runtime

Need to move the duplicated IsAsciiLetter method to just the source generator, since the source generator can't reference the char API. That way the runtime doesn't need to pay for a duplicated method.

* PR feedback.

Move helepr method to Stubs.cs.

src/libraries/System.Text.RegularExpressions/gen/RegexGenerator.Emitter.cs

@@ -4518,8 +4518,8 @@ private static string MatchCharacterClass(RegexOptions options, string chExpr, s
             // Next, handle simple sets of two ASCII letter ranges that are cased versions of each other, e.g. [A-Za-z].
             // This can be implemented as if it were a single range, with an additional bitwise operation.
             if (RegexCharClass.TryGetDoubleRange(charClass, out (char LowInclusive, char HighInclusive) rangeLower, out (char LowInclusive, char HighInclusive) rangeUpper) &&
-                RegexCharClass.IsAsciiLetter(rangeUpper.LowInclusive) &&
-                RegexCharClass.IsAsciiLetter(rangeUpper.HighInclusive) &&
+                CharExtensions.IsAsciiLetter(rangeUpper.LowInclusive) &&
+                CharExtensions.IsAsciiLetter(rangeUpper.HighInclusive) &&
                 (rangeLower.LowInclusive | 0x20) == rangeUpper.LowInclusive &&
                 (rangeLower.HighInclusive | 0x20) == rangeUpper.HighInclusive)
             {

src/libraries/System.Text.RegularExpressions/gen/Stubs.cs

@@ -58,6 +58,13 @@ public static int CommonPrefixLength(this ReadOnlySpan<char> span, ReadOnlySpan<
             return length;
         }
     }
+
+    internal static class CharExtensions
+    {
+        /// <summary>Gets whether the specified character is an ASCII letter.</summary>
+        public static bool IsAsciiLetter(char c) =>
+            (uint)((c | 0x20) - 'a') <= 'z' - 'a';
+    }
 }
 
 namespace System.Buffers

src/libraries/System.Text.RegularExpressions/src/System/Text/RegularExpressions/RegexCharClass.cs

@@ -990,10 +990,6 @@ public static bool IsAscii(ReadOnlySpan<char> s) // TODO https://github.com/dotn
             return true;
         }
 
-        /// <summary>Gets whether the specified character is an ASCII letter.</summary>
-        public static bool IsAsciiLetter(char c) =>
-            (uint)((c | 0x20) - 'a') <= 'z' - 'a';
-
         /// <summary>Gets whether we can iterate through the set list pairs in order to completely enumerate the set's contents.</summary>
         /// <remarks>This may enumerate negated characters if the set is negated.  This will return false if the set has subtraction.</remarks>
         private static bool CanEasilyEnumerateSetContents(string set) =>

src/libraries/System.Text.RegularExpressions/src/System/Text/RegularExpressions/RegexCompiler.cs

@@ -5443,8 +5443,8 @@ private void EmitMatchCharacterClass(string charClass)
             // Next, handle simple sets of two ASCII letter ranges that are cased versions of each other, e.g. [B-Db-d].
             // This can be implemented as if it were a single range, with an additional bitwise operation.
             if (RegexCharClass.TryGetDoubleRange(charClass, out (char LowInclusive, char HighInclusive) rangeLower, out (char LowInclusive, char HighInclusive) rangeUpper) &&
-                RegexCharClass.IsAsciiLetter(rangeUpper.LowInclusive) &&
-                RegexCharClass.IsAsciiLetter(rangeUpper.HighInclusive) &&
+                char.IsAsciiLetter(rangeUpper.LowInclusive) &&
+                char.IsAsciiLetter(rangeUpper.HighInclusive) &&
                 (rangeLower.LowInclusive | 0x20) == rangeUpper.LowInclusive &&
                 (rangeLower.HighInclusive | 0x20) == rangeUpper.HighInclusive)
             {