feat(discovery): enumerate the apps installed on the Shield (t57)

The live complaint ("it doesn't find all the apps") came from the curated
list: only four packages were ever offered, so anything the user had
installed elsewhere was invisible. The device knows its own apps.

ha_adb_common/installed_apps.py parses the two queries the sweep makes:

* `pm list packages -3` — every third-party (user-installed) package;
* `cmd package resolve-activity --brief -c android.intent.category.HOME`
  — the ACTIVE launcher, a system app that is nonetheless what the user
  opens, so it is appended rather than filtered away.

System apps are otherwise not enumerated (a Shield ships a few hundred and
they are not launchable apps); the parsers are tolerant of the noise a real
adb shell can prepend and of duplicate lines.

ShieldConnectionManager gains the sweep:
`async_refresh_installed_apps(force=False)` returns a cached
`[{package, name, icon}]` table, and `installed_apps()` is a pure read that
never issues a shell — so a state poll costs nothing.

Naming stays on the one path: the sweep calls `_dump_label`, the same
`pm dump` + parse + 24 h cache the entity path uses, now extracted so the
two cannot drift, including the t57 UNIT 0 rule that a fetch that returned
nothing is NOT cached. Labels are fetched APP_SWEEP_CONCURRENCY (4) at a
time with APP_SWEEP_SPACING (50 ms), so a typical Shield finishes in
seconds rather than the minutes a one-per-watchdog-cycle drain would take
while the socket still sees a trickle.

Failure is a normal outcome: no poller, a disconnected transport or a mute
shell stores an EMPTY table with a stamp and returns [], so nothing retries
in a loop and every surface degrades to "no discovered apps".

Icons: `pm dump` does NOT expose an app's icon resource — an extracted
APK's icon is a bitmap only its own code resolves — so this is honestly a
heuristic: APP_ICON_FALLBACKS (a curated package -> mdi: map) with
APP_ICON_GENERIC (mdi:television) for anything unknown, and always a string.

custom_components/nvidia_shield/const.py

@@ -151,6 +151,45 @@ DEFAULT_APP_DIRECTORY = {
     "com.google.android.youtube.tv": "YouTube",
 }
 
+# t57: best-effort icon per discovered app. `pm dump` does NOT expose the
+# app's icon resource (an extracted APK's icon is a bitmap only its own code
+# can resolve, and `dumpsys package` prints no drawable reference), so a device
+# icon is not obtainable over plain shell. These MDI names are therefore a
+# HEURISTIC: a curated package -> MDI mapping with a generic fallback, honest
+# about being a guess. The user's own `app_icon_map` (a picture URL or camera
+# entity, see CONF_APP_ICON_MAP) always outranks it for entity_picture.
+APP_ICON_FALLBACKS = {
+    "ar.tvplayer.tv": "mdi:television-classic",
+    "com.github.damontecres.wholphin": "mdi:jellyfish",
+    "com.netflix.ninja": "mdi:netflix",
+    "com.google.android.youtube.tv": "mdi:youtube-tv",
+    "com.google.android.tvlauncher": "mdi:view-dashboard",
+    "org.xbmc.kodi": "mdi:kodi",
+    "com.plexapp.android": "mdi:plex",
+    "com.amazon.amazonvideo.livingroom": "mdi:amazon",
+    "com.disney.disneyplus": "mdi:movie-open",
+    "com.spotify.tv.android": "mdi:spotify",
+    "com.google.android.apps.youtube.music": "mdi:music",
+    "com.valvesoftware.steamlink": "mdi:gamepad-variant",
+    "com.spotify.music": "mdi:spotify",
+}
+#: The fallback for an app this map has never heard of.
+APP_ICON_GENERIC = "mdi:television"
+
+# t57: installed-app discovery. The package list changes only when the user
+# installs or uninstalls something, so a long TTL is right; refreshing is
+# explicit (the options page and the app-list sensor ask for it) rather than a
+# per-poll cost.
+APP_LIST_TTL = 6 * 3600
+#: `pm dump` sweeps for labels run in a small burst — enough to finish a
+#: typical Shield quickly, few enough that the label fetches never queue
+#: behind each other on one socket or stall the watchdog.
+APP_SWEEP_CONCURRENCY = 4
+#: Spacing between the fetches of one sweep, per worker. Keeps the burst a
+#: trickle on the one socket the watchdog uses, while ~40 apps still finish in
+#: a few seconds instead of a few minutes.
+APP_SWEEP_SPACING = 0.05
+
 CHANNEL_DIGIT_DELAY = 0.1
 CONF_CHANNEL_DIGIT_DELAY = "channel_digit_delay"
 CONF_CHANNEL_SEND_ENTER = "channel_send_enter"