feat: HSAP platform v2 — modular navigation, quality review, audit log, world model simulation

Major changes:
- New frontend (platform/web/): Vite + React 18 + TypeScript + Tailwind
- 4-module navigation: 数据送标 / 模型管理 / 车队管理 / 系统管理
- Data catalog with charts (DMS/ADAS/Lane 3-tab view)
- Quality review workflow (标注质检): Good/Fine/Bad scoring with auto-advance
- Audit enhancements: batch operations, rejection categories, Feishu notifications
- Operation audit log (操作日志)
- World model simulation studio (仿真工坊)
- Dataset version management with snapshots and diff
- ADAS 7-class dataset integration (138K images organized + compressed)
- User management with Feishu integration and pagination
- CRUD/search/filter on all pages, card layout redesign
- PIL-optimized image overlay rendering
- Auto-snapshot on build, in_review workflow stage
- Removed embedded algorithm code (now in workspace)

platform/web/node_modules/nanoid/async/index.browser.cjs

@@ -0,0 +1,69 @@
+let random = async bytes => crypto.getRandomValues(new Uint8Array(bytes))
+
+let customAlphabet = (alphabet, defaultSize = 21) => {
+  // First, a bitmask is necessary to generate the ID. The bitmask makes bytes
+  // values closer to the alphabet size. The bitmask calculates the closest
+  // `2^31 - 1` number, which exceeds the alphabet size.
+  // For example, the bitmask for the alphabet size 30 is 31 (00011111).
+  // `Math.clz32` is not used, because it is not available in browsers.
+  let mask = (2 << (Math.log(alphabet.length - 1) / Math.LN2)) - 1
+  // Though, the bitmask solution is not perfect since the bytes exceeding
+  // the alphabet size are refused. Therefore, to reliably generate the ID,
+  // the random bytes redundancy has to be satisfied.
+
+  // Note: every hardware random generator call is performance expensive,
+  // because the system call for entropy collection takes a lot of time.
+  // So, to avoid additional system calls, extra bytes are requested in advance.
+
+  // Next, a step determines how many random bytes to generate.
+  // The number of random bytes gets decided upon the ID size, mask,
+  // alphabet size, and magic number 1.6 (using 1.6 peaks at performance
+  // according to benchmarks).
+
+  // `-~f => Math.ceil(f)` if f is a float
+  // `-~i => i + 1` if i is an integer
+  let step = -~((1.6 * mask * defaultSize) / alphabet.length)
+
+  return async (size = defaultSize) => {
+    let id = ''
+    while (true) {
+      let bytes = crypto.getRandomValues(new Uint8Array(step))
+      // A compact alternative for `for (var i = 0; i < step; i++)`.
+      let i = step | 0
+      while (i--) {
+        // Adding `|| ''` refuses a random byte that exceeds the alphabet size.
+        id += alphabet[bytes[i] & mask] || ''
+        if (id.length === size) return id
+      }
+    }
+  }
+}
+
+let nanoid = async (size = 21) => {
+  let id = ''
+  let bytes = crypto.getRandomValues(new Uint8Array((size |= 0)))
+
+  // A compact alternative for `for (var i = 0; i < step; i++)`.
+  while (size--) {
+    // It is incorrect to use bytes exceeding the alphabet size.
+    // The following mask reduces the random byte in the 0-255 value
+    // range to the 0-63 value range. Therefore, adding hacks, such
+    // as empty string fallback or magic numbers, is unneccessary because
+    // the bitmask trims bytes down to the alphabet size.
+    let byte = bytes[size] & 63
+    if (byte < 36) {
+      // `0-9a-z`
+      id += byte.toString(36)
+    } else if (byte < 62) {
+      // `A-Z`
+      id += (byte - 26).toString(36).toUpperCase()
+    } else if (byte < 63) {
+      id += '_'
+    } else {
+      id += '-'
+    }
+  }
+  return id
+}
+
+module.exports = { nanoid, customAlphabet, random }