第九轮(部分)：修复日报徽章配色、彻底重做PIN定位方案、清理静态文件品牌残留

- 日报卡片"已提交/未提交"状态徽章补色：DailyReportCard用<article><header><em>结构，原选择器
  要求.report-cards直接子<button>，日报卡片永远命中不了、只落到不分状态的兜底样式；月报因为
  行本身就是<button>所以没受影响。详情页.report-author em同理补了区分色（原来恒为绿色，"未
  提交"用绿色在语义上很怪）。
- 青少年模式/独立密码设置的PIN输入定位改用结构性方案，不再靠猜一个rem偏移量：给
  .auth-pin-boxes/.auth-pin-input包一层.auth-pin-wrap，输入框inset:0精确覆盖方块的真实
  渲染框，不管外层页面有没有标题、padding多少都自动对齐。上一轮的补丁（.youth-password-gate
  .auth-pin-input{top:1.75rem}）把重叠率从0%提到32.5%，本轮实测确认还是不够（方块下方
  67.5%的面积点不中），根治后4个采样点（四角+中心）全部精确命中输入框。过程中额外发现并
  修复了两个更隐蔽的层叠问题：全局input[type=text]的height:2.5rem特异度反超裸class、
  textarea/input共享的margin:0.25rem重置——不清掉这两个会让inset:0看起来生效了、实际尺寸/
  位置还是被悄悄改掉。
- .report-loading（举报原因页加载态）补基础样式，原来在base.css里零匹配，吃浏览器默认<p>样式。
- 清理静态文件里的Tinode品牌残留：index.html/index-dev.html的<meta>标签（application-name/
  description/og:*/canonical，此前只改了<title>，meta标签一直没跟上）、manifest.json（name/
  short_name/description改MeetingIM；prefer_related_applications+related_applications整个
  删掉——这两个字段会让Android/Chrome"添加到主屏幕"时引导用户去装无关的第三方开源Tinode App，
  不只是文案问题）；删除.well-known/assetlinks.json（声明了无关第三方App对本域名的URL处理/
  登录凭据代理权限，属于遗留模板残留，无真实对应关系，直接移除而非编造虚假内容）。

manifest.json里的图标文件(logo96.png/logo192.png/logo.svg)本身画的就是Tinode的蓝绿橙"t"标志，
是真实的视觉品牌泄漏，但重新设计一套图标涉及视觉识别决策，本次未动，留给用户决定。

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

css/base.css

@@ -1099,6 +1099,13 @@
   font-style: normal;
 }
 
+/* DailyReportCard renders <article><header><em>, not the <button> the two rules below (1196-1197)
+   assume -- so submitted/draft there always fell through to this rule's one fixed color, unable to
+   tell the two states apart (MonthlyReportList's own rows really are <button>s, so it never hit
+   this gap). Same colors as the button-based rules, just re-declared at matching structure. */
+.report-cards article > header em.submitted { color: #28784f; background: #e5f5ec; }
+.report-cards article > header em.draft { color: #ad6c11; background: #fff2dd; }
+
 .report-cards article p {
   display: -webkit-box;
   margin: 0.8rem 0;
@@ -1288,10 +1295,14 @@
 }
 
 .report-author em {
-  color: #2c8257;
   font-style: normal;
 }
 
+/* Was one fixed color (green) regardless of report.status's actual value -- a "未提交" (not yet
+   submitted) badge showing the same green as "已提交" reads backwards. */
+.report-author em.submitted { color: #28784f; }
+.report-author em.draft { color: #ad6c11; }
+
 .report-detail h2,
 .report-detail h3,
 .report-detail p {
@@ -6795,6 +6806,20 @@ div.toast.show {
   margin-top: 1.2rem;
 }
 
+/* .auth-pin-input used to be sized/positioned with hardcoded top/width/height values guessed to
+   match wherever .auth-pin-boxes happened to render on each page -- .auth-pin-content's <h1> and
+   .youth-password-gate's shorter top padding push the boxes to two different heights, so one
+   shared guessed `top` could only ever be right for one of them (confirmed live twice: the two
+   pages' real box positions differ by ~9.25rem, and a value tuned for one left the other's boxes
+   only ~32% overlapping the real input, still mostly unclickable). Wrapping both in their own
+   tightly-fitting, positioned container instead -- inset:0 on the input then always matches
+   .auth-pin-boxes' actual rendered box exactly, on any page, with no measured constant to keep
+   in sync. */
+.auth-pin-wrap {
+  position: relative;
+  display: inline-block;
+}
+
 .auth-pin-boxes {
   display: grid;
   grid-template-columns: repeat(4, 4rem);
@@ -6812,21 +6837,19 @@ div.toast.show {
 
 .auth-pin-input {
   position: absolute;
-  top: 11rem;
-  width: 19rem;
-  height: 5rem;
+  inset: 0;
+  margin: 0;
   opacity: 0;
 }
 
-/* .auth-pin-input's shared top:11rem was measured against .auth-pin-content's layout (6rem
-   top padding + an <h1> pushing .auth-pin-boxes further down) -- .youth-password-gate has no
-   title and only 3rem of top padding, so .auth-pin-boxes sits ~9.25rem higher here and the
-   invisible real input ended up entirely below the visible boxes (confirmed live: boxes at
-   86.7-140px, input at 210-243px, a real click on the boxes hit nothing and never refocused
-   the input). Re-anchored to this container's own shorter layout instead of reusing the other
-   page's absolute value. */
-.youth-password-gate .auth-pin-input {
-  top: 1.75rem;
+/* input[type="text"]'s global reset (2131 above) sets a fixed height:2.5rem at higher specificity
+   (0,1,1) than a bare .auth-pin-input's (0,1,0) -- it was winning and pinning this input to 2.5rem
+   tall regardless of the boxes' actual height, undoing the inset:0 fix above for the vertical axis
+   (width happened to still come out right only because the reset's own width:100% agrees with this
+   element's actual need). Selector bumped to definitively outrank it, not just happen to. */
+.auth-pin-wrap .auth-pin-input {
+  width: 100%;
+  height: 100%;
 }
 
 .auth-pin-content .auth-primary {
@@ -14061,6 +14084,13 @@ div.toast.show {
   font-weight: 500;
 }
 
+.report-loading {
+  margin: 0;
+  padding: 1rem;
+  color: #6e7781;
+  font-size: 0.85rem;
+}
+
 .report-reason-list {
   background: #fff;
 }