Reimplement resolveExtractor() as wrapper over resolveLanguages()

By wrapping `resolveLanguages()`, which is memoized, we can avoid executing `codeql resolve extractor` several times over the course of an analysis.

src/codeql.ts

@@ -946,32 +946,11 @@ async function getCodeQLForCmd(
       await new toolrunner.ToolRunner(cmd, args).exec();
     },
     async resolveExtractor(language: Language): Promise<string> {
-      // Request it using `format=json` so we don't need to strip the trailing new line generated by
-      // the CLI.
-      let extractorPath = "";
-      await new toolrunner.ToolRunner(
-        cmd,
-        [
-          "resolve",
-          "extractor",
-          "--format=json",
-          `--language=${language}`,
-          "--extractor-include-aliases",
-          ...getExtraOptionsFromEnv(["resolve", "extractor"]),
-        ],
-        {
-          silent: true,
-          listeners: {
-            stdout: (data) => {
-              extractorPath += data.toString();
-            },
-            stderr: (data) => {
-              process.stderr.write(data);
-            },
-          },
-        },
-      ).exec();
-      return JSON.parse(extractorPath) as string;
+      // A previous implementation executed `codeql resolve extractor`.
+      // This can be a bit slow due to the JVM startup cost. Instead, get
+      // the extractor path from resolveLanguages(), which caches its output.
+      const extractors = await this.resolveLanguages();
+      return extractors[language][0];
     },
     async resolveQueriesStartingPacks(queries: string[]): Promise<string[]> {
       const codeqlArgs = [