Since .st files are preprocessed with cpp, SNL code may be included. This
patch adds rules to create .st.d files using the mkmf.pl tool from base.
Also removed some obsolete stuff and fixed the CLEANS definition.

Silently changing the meaning of a user program is not nice. Besides,
using the implicit default queue size is deprecated anyway.

This partly a preparation for proper multi-PV array support.

Since we now resolve built-in functions as well as constants during
analysis, we no longer need to access the symbol table in the code
generator. This means we can get rid of an ugly hack. The gen_ss_init
procedure is also no longer needed and has been removed.

The problem was a call seq_monitor(ch, FALSE) inside pvAssign after we
destroy the channel (if one exists). This will always fail (strange I
never noticed that) and as a result dbch->monid would not be reset to
NULL. That in turn means when the channel is re-assigned and connects,
seq_monitor(ch, TRUE) will not do anything because it thinks there is
already a valid subscription.

The fix is to remove the call and instead only reset dbch->monid to NULL.

The function seq_monitor is called by user code via pvMonitor and
pvStopMonitor as well as from the CA connection handler. So we should
make sure that there cannot be a race between the two.

This did not solve the original problem reported by Janez Golob
<janez.golob@cosylab.com>, but it's a valid bugfix nevertheless.

The first removed paragraph was just wrong. The second one proposed
alternatives which I no longer regard as being superior. Connection loss
is not supposed to happen regularly during normal operation, so the
small overhead of waking all state sets is negligable in practice.

We give the hint to use option +p only if the function actually supports
multi-PV arrays. We know this by examining the add_length column in the
builtin function table (functions that support them must add the array
length as implicit argument to the call).

Also fixed/extended pvSync.st by adding an index to the argument of the
pvPut call.

Foreign declarations have been a bad idea. This became clear after
implementing indirect calls, because as a consequence you get 'used but
not declared' warnings for calls to foreign functions, too. I could
'fix' that with a bit of effort, but that would be misplaced. The C
compiler will warn about undeclared variables and functions anyway, so
why bother? If you really want to know, perhaps for debugging, use -W
option for extra warnings. If you do, you'll now get the more precise
waring 'treating undeclared object xxx as foreign'.

Original comment:

  In the lemon parser generator, change all hashes to unsigned to avoid
  potential problems with signed integer overflow.