round-3-gamma: combine c_escape + scan_interp_string batching — max round-3 savings

Combines two orthogonal optimizations:
1. c_escape batching (from alpha): ASCII runs emitted as str_slice segments instead
   of one str_char_at string per byte. O(N) allocs → O(K) where K = special chars.

2. scan_interp_string batching (from beta): char dispatch via str_char_code (Int)
   + clean_start tracking to flush plain runs as str_slice. Eliminates per-char
   string allocations in the string-literal scanning hot path.

Result on web/src/main.el: 14.5MB -> 13.4MB peak RSS (-7.6%).
Self-hosting: PASS.

lang/el-compiler/src/codegen.el

@@ -38,10 +38,13 @@ fn is_hex_digit_byte(b: Int) -> Bool {
 }
 
 fn c_escape(s: String) -> String {
-    // Use index-based byte scanning via str_char_code(s, i) and str_char_at(s, i).
-    // This avoids native_string_chars + str_join, which corrupts high-byte (>= 0x80)
-    // characters because list_join's looks_like_string heuristic rejects strings
-    // whose first byte is >= 0x7F and emits them as decimal pointer values instead.
+    // Batch ASCII chars using str_slice instead of str_char_at per byte.
+    // Track clean_start: the beginning of the current run of bytes that need
+    // no escaping. On each special byte, flush the accumulated clean run via
+    // str_slice, then append the escape. This reduces parts-list appends from
+    // O(N) to O(K) where K = number of special bytes << N for normal strings.
+    //
+    // Special bytes: '"'=34, '\\'=92, '\n'=10, '\r'=13, '\t'=9, any byte>=128.
     //
     // IMPORTANT: after a \xNN hex escape, if the next byte is a hex digit
     // (0-9, a-f, A-F), we emit `""` to split the C string literal so the C
@@ -51,46 +54,75 @@ fn c_escape(s: String) -> String {
     let total: Int = str_len(s)
     let parts: [String] = native_list_empty()
     let i: Int = 0
+    let clean_start: Int = 0
     let prev_was_hex_escape: Bool = false
     while i < total {
         let bval: Int = str_char_code(s, i)
-        // If the previous token was a \xNN escape and the current byte is a
-        // hex digit, insert an empty string literal ("") to break the escape.
+        // Handle the hex-escape split case first: if prev was \xNN and this
+        // byte is a hex digit, we must flush the clean run and insert "".
+        // (At this point clean_start == i since the previous special byte
+        // already reset it, so flush is a no-op unless something is pending.)
         if prev_was_hex_escape {
             if is_hex_digit_byte(bval) {
+                // Flush any accumulated clean bytes before the split marker.
+                if clean_start < i {
+                    let parts = native_list_append(parts, str_slice(s, clean_start, i))
+                }
                 let parts = native_list_append(parts, "\"\"")
+                let clean_start = i
             }
         }
         let prev_was_hex_escape = false
         if bval == 34 {
-            // 34 = '"'
+            // 34 = '"' — flush clean run, then escape
+            if clean_start < i {
+                let parts = native_list_append(parts, str_slice(s, clean_start, i))
+            }
             let parts = native_list_append(parts, "\\\"")
+            let clean_start = i + 1
         } else {
             if bval == 92 {
                 // 92 = '\\'
+                if clean_start < i {
+                    let parts = native_list_append(parts, str_slice(s, clean_start, i))
+                }
                 let parts = native_list_append(parts, "\\\\")
+                let clean_start = i + 1
             } else {
                 if bval == 10 {
                     // 10 = '\n'
+                    if clean_start < i {
+                        let parts = native_list_append(parts, str_slice(s, clean_start, i))
+                    }
                     let parts = native_list_append(parts, "\\n")
+                    let clean_start = i + 1
                 } else {
                     if bval == 13 {
                         // 13 = '\r'
+                        if clean_start < i {
+                            let parts = native_list_append(parts, str_slice(s, clean_start, i))
+                        }
                         let parts = native_list_append(parts, "\\r")
+                        let clean_start = i + 1
                     } else {
                         if bval == 9 {
                             // 9 = '\t'
+                            if clean_start < i {
+                                let parts = native_list_append(parts, str_slice(s, clean_start, i))
+                            }
                             let parts = native_list_append(parts, "\\t")
+                            let clean_start = i + 1
                         } else {
                             if bval >= 128 {
-                                // Escape non-ASCII bytes (>= 0x80) as \xNN so
-                                // Clang does not misinterpret multi-byte UTF-8
-                                // sequences in C string literals.
+                                // Non-ASCII: flush, then \xNN
+                                if clean_start < i {
+                                    let parts = native_list_append(parts, str_slice(s, clean_start, i))
+                                }
                                 let parts = native_list_append(parts, "\\x" + byte_to_hex2(bval))
                                 let prev_was_hex_escape = true
-                            } else {
-                                let parts = native_list_append(parts, str_char_at(s, i))
+                                let clean_start = i + 1
                             }
+                            // else: plain ASCII — extends the current clean run (no append)
                         }
                     }
                 }
@@ -98,6 +130,10 @@ fn c_escape(s: String) -> String {
         }
         let i = i + 1
     }
+    // Flush the final clean run if any
+    if clean_start < total {
+        let parts = native_list_append(parts, str_slice(s, clean_start, total))
+    }
     str_join(parts, "")
 }