Use named capture groups for version string matching

Previously it was referencing the wrong group index (off by one), which is easy to do when relying on indices. Using capture groups here should make it less error prone (and also fixes this initial problem).

Long term a small unit test should avoid this happening / regressing.

common.js

@@ -179,13 +179,13 @@ function versionLessThan(cur_ver, min_ver) {
 // 'dev' is `null` if `str` was not a dev version.
 // On failure, returns `null`.
 function parseVersion(str) {
-  const match = /^(\d+)\.(\d+)\.(\d+)(?:-dev\.(\d+)\+[0-9a-f]*)?$/.exec(str);
+  const match = /^(?<major>\d+)\.(?<minor>\d+)\.(?<patch>\d+)(?:-dev\.(?<dev>\d+)\+[0-9a-f]*)?$/.exec(str);
   if (match === null) return null;
   return {
-    major: parseInt(match[0]),
-    minor: parseInt(match[1]),
-    patch: parseInt(match[2]),
-    dev: match[3] === null ? null : parseInt(match[3]),
+    major: parseInt(match.groups['major']),
+    minor: parseInt(match.groups['minor']),
+    patch: parseInt(match.groups['patch']),
+    dev: match.groups['dev'] === undefined ? null : parseInt(match.groups['dev']),
   };
 }