feat: add tokenReducer, vitest tests, fix merge label convention

src/utils/tokenReducer.js

@@ -0,0 +1,36 @@
+// Reduces consecutive same-face moves into their net rotation.
+// Agnostic to move names — works with any single-letter move notation.
+
+const EMPTY = 'E';
+const MODS = [EMPTY, '', '2', "'"];
+
+const reduceGroup = (group) => {
+    const sum = group.reduce((acc, curr) => acc + MODS.indexOf(curr.mod), 0);
+    const mod = MODS[sum % 4];
+    return mod === EMPTY ? '' : `${group[0].name}${mod}`;
+};
+
+export const parseToken = (token) => {
+    const match = token.match(/^(\w)(.?)$/);
+    if (!match) return null;
+    return { token, name: match[1], mod: match[2] };
+};
+
+export const tokenReducer = (tokens) => {
+    const parsed = tokens.map(parseToken).filter(Boolean);
+    const desc = [];
+    const res = [];
+    let lastPos = 0;
+
+    for (let i = 0; i <= parsed.length; i++) {
+        if (i === parsed.length || (i > lastPos && parsed[i].name !== parsed[lastPos].name)) {
+            const group = parsed.slice(lastPos, i);
+            const reduced = reduceGroup(group);
+            desc.push({ reduced, group });
+            if (reduced !== '') res.push(reduced);
+            lastPos = i;
+        }
+    }
+
+    return { desc, tokens: res };
+};