Reduce routine RX power-saving noise-floor refreshes

docs/test-results/radio_receive_calibration.md

@@ -11,6 +11,29 @@ Periodic requests coalesce while a block is active. RXPS resumes after a complet
 block, including a held block, or at timeout. Receive and transmit activity are
 never interrupted just to finish calibration.
 
+In single-profile RXPS, routine refreshes wait at least 60 seconds after the
+last completed or timed-out block before requesting another continuous-RX
+window. Both periodic maintenance and passive retry checks use this cooldown;
+retries do not restart a partially collected block. With RXPS off, or when the
+selected tuple requires continuous-RX fallback, the existing two-second
+maintenance tick and passive-probe age check are unchanged. Startup, explicit
+gain/tuning changes, AGC resets, and radio recovery still request a fresh block
+promptly. Dual-profile scanning retains
+its separate refresh policy.
+
+Frequent configured AGC resets can still dominate calibration time: Cascade
+repeater builds default to an eight-second AGC interval, and MQTT observers
+use 28 seconds. Those resets are not throttled by the routine cooldown. After
+a timed-out startup block with no valid floor, the next routine RXPS attempt
+also waits 60 seconds; passive retries retain their existing bounded busy
+escape while waiting for a valid baseline.
+
+The longer RXPS cooldown trades a slower response to ambient noise changes for
+more time in duty-cycle RX. It does not reduce sample count, shorten settling,
+or put the radio to sleep between samples. A large upward change still needs
+three complete high blocks to pass the rise hold. No battery-life improvement
+has been measured for this scheduling change.
+
 The block estimate is the 12.5th percentile: the average of the eighth and
 ninth sorted samples, retaining fractional dBm. A published update uses 25% of
 the previous floor and 75% of the new estimate, with a -120 dBm lower bound. Initial calibration and explicit gain/tuning changes seed a new baseline.
@@ -30,9 +53,10 @@ out of 64 to avoid higher-valued traffic entirely. Nearly continuous
 interference or enough low outliers can still mislead it. The rise hold delays
 large changes but cannot identify their source.
 
-RSSI sampling waits 20 ms after every receive re-arm, including CAD and TX
-completion, to avoid unsettled frontend readings. The ten-second trial window
-often failed to complete during heavy traffic with that settling guard; the
+RSSI sampling waits 7 ms on SX1262, SX1268, and LLCC68 builds, or 20 ms on other
+families, after every receive re-arm, including CAD and TX completion, to avoid
+unsettled frontend readings. The ten-second trial window often failed to
+complete during heavy traffic with that settling guard; the
 30-second cap allows more idle opportunities. Partial blocks still time out
 without publishing. Longer collection spends more time in continuous RX when
 RXPS is enabled. A quiet minimum is about 3.2 seconds; real completion time