fix(admin): fix rescan goroutine lifecycle and race on shared ScanProgress

- Protect adminService scan state (cancel func + progress pointer) with sync.Mutex.
- Allocate fresh ScanProgress per trigger and pass it to the scanner,
  eliminating races on the previously shared progress struct.
- Cancel previous scan context before starting a new one.
- Add tests for cancellation, fresh progress per scan, concurrent triggers,
  and empty progress when never started.
- Fix race-prone tests by polling Running==true before waiting for completion.

1 files changed, 21 insertions, 0 deletions

diff --git a/web/css/player.css b/web/css/player.css
index c6070f9..0f468f9 100644
--- a/web/css/player.css
+++ b/web/css/player.css
@@ -137,6 +137,27 @@
   background: var(--player-stage-bg);
 }
 
+/* Image viewer inside player */
+.player .media-image {
+  width: 100%;
+  height: 100%;
+  object-fit: contain;
+  display: block;
+  transition: transform 0.15s ease;
+  user-select: none;
+}
+.player .media-image.hidden {
+  display: none;
+}
+
+/* Show/hide controls per media type */
+.player.has-image .video-only {
+  display: none !important;
+}
+.player:not(.has-image) .image-only {
+  display: none !important;
+}
+
 /* Overlay controls */
 .player .controls {
   flex: 0 0 auto;