etc: add redfishpower tuning notes

Problem: On redfish systems where the on/off time can take a very
long time, the default "wait until" retry delay of 1 second can
lead to an excessive number of requests to be sent out.  Users would
not be aware of this and how it can be tuned.

Add some additional comments to all redfishpower device files to inform
users of this potential tweak.  Have the "wait until delay" configured
in the loginc script section so it can be easily tuned by the user.

etc/redfishpower-cray-r272z30.dev

@@ -40,6 +40,16 @@
 #   Note that if "cycle_ranged" is scripted as an "off" followed by an
 #   "on", the timeout should account for the combined time.
 #
+#   D) The "wait until delay" indicates how often redfishpower should
+#   check if a node is confirmed to be "on"/"off".  It defaults to 1
+#   second.
+#
+#   On systems where the time to confirm "on" / "off" is long, this may
+#   lead to an excess and unnecessary amount of extra traffic.  This is
+#   especially true at scale.
+#
+#   This can be adjusted to something larger in the logic section below.
+#
 specification "redfishpower-cray-r272z30" {
 	timeout 	60
 
@@ -59,6 +69,8 @@ specification "redfishpower-cray-r272z30" {
 		expect "redfishpower> "
 		send "settimeout 60\n"
 		expect "redfishpower> "
+		send "setwaituntildelay 1\n"
+		expect "redfishpower> "
 	}
 	script logout {
 		send "quit\n"

etc/redfishpower-cray-windom.dev

@@ -47,6 +47,16 @@
 #   Note that if "cycle_ranged" is scripted as an "off" followed by an
 #   "on", the timeout should account for the combined time.
 #
+#   D) The "wait until delay" indicates how often redfishpower should
+#   check if a node is confirmed to be "on"/"off".  It defaults to 1
+#   second.
+#
+#   On systems where the time to confirm "on" / "off" is long, this may
+#   lead to an excess and unnecessary amount of extra traffic.  This is
+#   especially true at scale.
+#
+#   This can be adjusted to something larger in the logic section below.
+#
 specification "redfishpower-cray-windom-node0" {
 	timeout 	60
 
@@ -64,6 +74,8 @@ specification "redfishpower-cray-windom-node0" {
 		expect "redfishpower> "
 		send "settimeout 60\n"
 		expect "redfishpower> "
+		send "setwaituntildelay 1\n"
+		expect "redfishpower> "
 	}
 	script logout {
 		send "quit\n"
@@ -109,6 +121,8 @@ specification "redfishpower-cray-windom-node1" {
 		expect "redfishpower> "
 		send "settimeout 60\n"
 		expect "redfishpower> "
+		send "setwaituntildelay 1\n"
+		expect "redfishpower> "
 	}
 	script logout {
 		send "quit\n"

etc/redfishpower-supermicro.dev

@@ -40,6 +40,16 @@
 #   Note that if "cycle_ranged" is scripted as an "off" followed by an
 #   "on", the timeout should account for the combined time.
 #
+#   D) The "wait until delay" indicates how often redfishpower should
+#   check if a node is confirmed to be "on"/"off".  It defaults to 1
+#   second.
+#
+#   On systems where the time to confirm "on" / "off" is long, this may
+#   lead to an excess and unnecessary amount of extra traffic.  This is
+#   especially true at scale.
+#
+#   This can be adjusted to something larger in the logic section below.
+#
 specification "redfishpower-supermicro" {
 	timeout 	60
 
@@ -59,6 +69,8 @@ specification "redfishpower-supermicro" {
 		expect "redfishpower> "
 		send "settimeout 60\n"
 		expect "redfishpower> "
+		send "setwaituntildelay 1\n"
+		expect "redfishpower> "
 	}
 	script logout {
 		send "quit\n"