Colour list markers by depth, highlight fences, draw images as links, and stream a list an item at a time

Items 1-5 of TRANSCRIPT_RENDERING.md's list, plus the AGP 9.4.0 bump from 7.
MarkdownRoot provides the renderer's locals itself instead of calling its
Markdown() composable; fences and indented blocks go through CodeFence.kt,
which shares the tool-input highlighter and a fence-language alias table;
an image in a paragraph is a link carrying its alt text, so every paragraph
is now platform text; LiveParse freezes the finished items of the tail list
so a forty-item list streams as forty paragraphs would.

Measured before, on the emulator (report from transcript-bench.sh over the
200-line fence fixture): draw phase 0.72ms per frame, transcript 0.36ms.
stream-bench.sh (new) streaming forty linked bullets on the old build:
markdown reparsed while streaming 483, 3.9ms mean, 11.6ms worst; record:
one block worst 1.6ms. The after runs, the on-screen check of the glyphs
and lint are recorded as owed in the doc's "What is next".

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

app/stream-bench.sh

@@ -0,0 +1,59 @@
+#!/bin/sh
+# Streams a message into the sandbox session on screen and prints the app's
+# render report for the time it took -- the standard measurement for "is a
+# reply that is still arriving cheap", the way transcript-bench.sh is for
+# scrolling one that has settled.
+#
+#   ./stream-bench.sh FILE           restart the app, open the first
+#                                    session, stream FILE into it, report
+#   ./stream-bench.sh -k FILE        keep whatever session is open now
+#
+# The session is the first one the sandbox lists (ui-sandbox.sh spawn makes
+# one), and the file is echoed back a word at a time, which is the shape a
+# real reply arrives in. The numbers to read are `markdown reparsed while
+# streaming` -- how many times, and how long each -- and the worst
+# `record: one block`; both are proportional to how much of the reply the
+# tail reparse has to cover, which is what the live parse exists to bound.
+set -eu
+cd "$(dirname "$0")"
+. ./android-env.sh >/dev/null 2>&1
+
+keep=""
+while getopts k flag; do
+    case $flag in
+    k) keep=1 ;;
+    *) exit 2 ;;
+    esac
+done
+shift $((OPTIND - 1))
+file=${1:?usage: stream-bench.sh [-k] FILE}
+
+sid=$(./ui-sandbox.sh api /sessions | python3 -c 'import json,sys; print(json.load(sys.stdin)[0]["id"])')
+
+if [ -z "$keep" ]; then
+    adb shell am force-stop com.example.aiapp
+    adb shell am start -n com.example.aiapp/.MainActivity >/dev/null
+    sleep 5
+    ui-trace record -d 3000 --do 'tap 500 545' -o /tmp/bench-open.txt >/dev/null 2>&1
+    sleep 3
+fi
+
+# The first tap resets the report's window; see transcript-bench.sh.
+ui-trace record -d 2000 --do 'tap 723 205' -o /tmp/bench-reset.txt >/dev/null 2>&1
+adb logcat -c
+
+./ui-sandbox.sh send "$sid" "@$file" >/dev/null
+# Until the echo has finished: its status goes back to idle.
+i=0
+while [ "$i" -lt 120 ]; do
+    status=$(./ui-sandbox.sh api "/sessions/$sid" |
+        python3 -c 'import json,sys; print(json.load(sys.stdin)["status"])')
+    [ "$status" = idle ] && break
+    sleep 2
+    i=$((i + 1))
+done
+sleep 1
+
+ui-trace record -d 2000 --do 'tap 723 205' -o /tmp/bench-report.txt >/dev/null 2>&1
+sleep 1
+adb logcat -d -s ai-app:I | sed -n '/ai-app render report/,$p' | sed 's/^.*ai-app  : //'