Faster tokenizer lookahead (#13341)

* refactor: simplify token context structure

* add benchmark

* perf: return a sub-state on tokenizer lookahead

* Update packages/babel-parser/src/tokenizer/index.js

Co-authored-by: Brian Ng <bng412@gmail.com>

* Update packages/babel-parser/src/tokenizer/index.js

Co-authored-by: Brian Ng <bng412@gmail.com>

* remove irrelevant comment

* fix: guard curPosition with isLookahead

* add test cases

Co-authored-by: Brian Ng <bng412@gmail.com>

packages/babel-parser/benchmark/many-arrow-function-flow-types/bench.mjs

@@ -0,0 +1,22 @@
+import Benchmark from "benchmark";
+import baseline from "@babel-baseline/parser";
+import current from "../../lib/index.js";
+import { report } from "../util.mjs";
+
+const suite = new Benchmark.Suite();
+function createInput(length) {
+  return "type A = " + "| (x) => void".repeat(length);
+}
+function benchCases(name, implementation, options) {
+  for (const length of [256, 512, 1024, 2048]) {
+    const input = createInput(length);
+    suite.add(`${name} ${length} arrow function types`, () => {
+      implementation.parse(input, options);
+    });
+  }
+}
+
+benchCases("baseline", baseline, { plugins: ["flow"] });
+benchCases("current", current, { plugins: ["flow"] });
+
+suite.on("cycle", report).run();