Fixup for 17b665d: remove 'sleep' hack & fix race condition in tests

17b665d was a failed attempt to hack around a race condition by
lengthening a 'sleep'; this simply made the problem appear less
frequently.

The problem only occurred in the toolchanger/toolno-pocket-differ/*
tests.  Comparing with other successful linuxcncrsh-based tests turned
up 9748fe9 and the "set wait done" command, which clears up the
problem, even with the original 'sleep' hack removed.

Author: zultron

scripts/linuxcnc.in

@@ -600,8 +600,6 @@ if ! program_available $EMCSERVER; then
 fi
 export INI_FILE_NAME="$INIFILE"
 $EMCSERVER -ini "$INIFILE" &
-# give linuxcnc *five* seconds to finish:  race condition alert!
-sleep 5
 
 # 4.3.2. Start REALTIME
 echo "Loading Real Time OS, RTAPI, and HAL_LIB modules" >>$PRINT_FILE

tests/toolchanger/toolno-pocket-differ/shared-test.sh

@@ -44,6 +44,7 @@ fi
         echo 'set mdi m100 P3 Q#5400'             # toolno
         echo 'set mdi m100 P4 Q#5403'             # TLO z
         echo 'set mdi m100 P5'                    # blank line
+        echo "set wait done"
     }
 
     echo hello EMC mt 1.0
@@ -150,10 +151,6 @@ fi
     echo set mdi g43
     introspect 30
 
-
-    # give linuxcnc a second to finish
-    sleep 1.0
-
     echo shutdown
 ) | nc localhost 5007