traces: show the stack after reading a token

Currently, if we have long rules and series of shift, we stack states
without showing stack.  Let's be more incremental, and do how the Java
skeleton does.

* data/skeletons/lalr1.cc, data/skeletons/lalr1.d,
* data/skeletons/yacc.c:
Here.
Adjust test cases.
* tests/torture.at (AT_DATA_STACK_TORTURE): Disable stack traces: this
test produces a very large stack, and showing the stack each time we
shift a token goes quadatric.

tests/c++.at

@@ -648,25 +648,30 @@ AT_FULL_COMPILE([[input]])
 AT_PARSER_CHECK([[input]], [[0]], [[]],
 [[Starting parse
 Entering state 0
+Stack now 0
 Reading a token
 Next token is token NUMBER (1)
 Shifting token NUMBER (1)
 Entering state 1
+Stack now 0 1
 Reducing stack by rule 1 (line 34):
    $1 = token NUMBER (1)
 -> $$ = nterm expr (10)
 destroy: 1
 Stack now 0
 Entering state 2
+Stack now 0 2
 Reading a token
 Next token is token NUMBER (30)
 Reducing stack by rule 2 (line 35):
 -> $$ = nterm @1 (20)
 Stack now 0 2
 Entering state 4
+Stack now 0 2 4
 Next token is token NUMBER (30)
 Shifting token NUMBER (30)
 Entering state 5
+Stack now 0 2 4 5
 Reducing stack by rule 3 (line 35):
    $1 = nterm expr (10)
    $2 = nterm @1 (20)
@@ -678,11 +683,13 @@ destroy: 20
 destroy: 10
 Stack now 0
 Entering state 2
+Stack now 0 2
 Reading a token
 Next token is token EOI ()
 Shifting token EOI ()
 Entering state 3
 Stack now 0 2 3
+Stack now 0 2 3
 Cleanup: popping token EOI ()
 Cleanup: popping nterm expr (40)
 destroy: 40