Add strict mode and template literal error handling for unrecognized escape sequences

This commit addresses the PR review feedback to properly handle unrecognized
escape sequences according to ECMAScript specification:

- In template literals: unrecognized escape sequences are always syntax errors
- In strict mode: unrecognized escape sequences are syntax errors
- In non-strict mode: unrecognized escape sequences preserve the backslash

The implementation follows the same pattern used for octal escape sequences,
ensuring consistent error handling behavior across the lexer.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: github-actions[bot]

crates/swc_ecma_lexer/src/common/lexer/mod.rs

@@ -1563,7 +1563,18 @@ pub trait Lexer<'a, TokenAndSpan>: Tokens<TokenAndSpan> + Sized {
             // the following character. According to ECMAScript, when a backslash precedes
             // a character that doesn't form a valid escape sequence, both the backslash
             // and the character should be preserved in the string value.
+            //
+            // However, in strict mode, unrecognized escape sequences are syntax errors.
+            // In template literals, they should always be errors (pre-ES2018 behavior).
             _ => {
+                // In template literals, unrecognized escape sequences are always errors
+                if in_template {
+                    self.error(start, SyntaxError::InvalidStrEscape)?
+                }
+
+                // In strict mode, unrecognized escape sequences are syntax errors
+                self.emit_strict_mode_error(start, SyntaxError::InvalidStrEscape);
+
                 // Don't bump - let the following character be read normally in the next
                 // iteration
                 return Ok(Some(CodePoint::from_char('\\')));