[PERF] tokenizer: avoid regexp match

When a matching a string against a regexp, some memory is allocated.
	tokenize	tokenizeSymbol	Model
Before	422ms		203ms		3382ms
	215Mb		127Mb

After	407ms		188ms		3335ms
	(-3.4%)		(-7.2%)		(-1.38%)
	197Mb		106Mb
	(-8.3%)		(-16.5%)

(averages over 5 runs)

closes #4191

Task: 3922347
Signed-off-by: Pierre Rousseau (pro) <pro@odoo.com>

src/formulas/tokenizer.ts

@@ -143,7 +143,7 @@ function tokenizeString(chars: TokenizingChars): Token | null {
   return null;
 }
 
-const separatorRegexp = /^[\w\.!\$]+/;
+const SYMBOL_CHARS = new Set("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_.!$");
 
 /**
  * A "Symbol" is just basically any word-like element that can appear in a
@@ -184,11 +184,8 @@ function tokenizeSymbol(chars: TokenizingChars): Token | null {
       };
     }
   }
-  const match = chars.remaining().match(separatorRegexp);
-  if (match) {
-    const value = match[0];
-    result += value;
-    chars.advanceBy(value.length);
+  while (chars.current && SYMBOL_CHARS.has(chars.current)) {
+    result += chars.shift();
   }
   if (result.length) {
     const value = result;