fix(perf): Pulsar stays fast in every mode — the editor IS the instrument

perf-audio deprioritized Pulsar to nice 10 in cool/extreme and nice 5 in
standard, on the theory that the editor should yield to audio. That trade buys
nothing. The entire audio chain runs SCHED_FIFO *above* it — scsynth 90,
sclang 85, PipeWire 95, WirePlumber 95, Ardour 80 — and a realtime policy
preempts every CFS nice level unconditionally. So Pulsar at -5 cannot take a
single cycle from SuperDirt; it only wins against ordinary processes (browsers,
indexers, builds), which is exactly what we want during a composing session.

Now nice -5 in every mode, overridable with PULSAR_NICE=NN (0 = neutral,
positive = the old behaviour). -5 rather than 0 because the renderer is a single
JS thread and its work is bursty: it has to win the CPU on the tick after a
keystroke, then yield. No root needed — @audio's `nice -19` grant in
/etc/security/limits.conf is what makes the same renice work for pln.

Validated: installed to /usr/local/sbin/perf-audio, re-applied via
`bridge.py perf standard`, all Pulsar processes confirmed at nice -5 live.

perf.sh

@@ -659,21 +659,26 @@ set_priorities() {
   fi
 
   if [ ! -z "$PULSAR_PIDS" ]; then
-    if [ "$mode" != "standard" ]; then
-      # Extreme/cool: low but NOT idle — keeps editor usable while livecoding
-      for PID in $PULSAR_PIDS; do
-        renice -n 10 -p $PID >/dev/null 2>&1
-        ionice -c 2 -n 6 -p $PID >/dev/null 2>&1
-      done
-      echo "✓ Set Pulsar processes to low priority (nice 10, best-effort I/O 6)"
-    else
-      # Standard mode: slightly lower
-      for PID in $PULSAR_PIDS; do
-        renice -n 5 -p $PID >/dev/null 2>&1
-        ionice -c 2 -n 5 -p $PID >/dev/null 2>&1
-      done
-      echo "✓ Set Pulsar processes to slightly lower priority (nice 5)"
-    fi
+    # Pulsar is the INSTRUMENT, not background load, so it stays fast in EVERY
+    # mode. The old policy (nice 10 in cool/extreme, nice 5 in standard) cost
+    # real editor latency and bought no audio headroom, because the entire audio
+    # chain runs SCHED_FIFO above it: scsynth 90, sclang 85, PipeWire 95,
+    # WirePlumber 95, Ardour 80. A realtime policy preempts EVERY CFS nice
+    # level, so Pulsar at -5 cannot take a single cycle from SuperDirt — it only
+    # wins against other ordinary processes (browsers, indexers, builds), which
+    # is exactly what we want during a composing session.
+    #
+    # -5 rather than 0 because the renderer is single-threaded and its work is
+    # bursty: it must win the CPU on the very next tick after a keystroke, then
+    # yield. Needs @audio's `nice -19` grant in /etc/security/limits.conf —
+    # which is also why the same renice works for pln without root.
+    #
+    # Override with PULSAR_NICE=NN (0 = neutral, positive = the old behaviour).
+    for PID in $PULSAR_PIDS; do
+      renice -n "${PULSAR_NICE:--5}" -p $PID >/dev/null 2>&1
+      ionice -c 2 -n 2 -p $PID >/dev/null 2>&1
+    done
+    echo "✓ Set Pulsar to HIGH priority (nice ${PULSAR_NICE:--5}, I/O best-effort 2) — editor stays reactive"
   else
     echo "ℹ️ Pulsar processes not found - skipping"
   fi