Instead, we use teh same binary we produce for the test case, and
add an extra switch -t that enables running the sequencer program.

If epicsEventTryWait succeeds, there could still be a previous request
pending that has timed out. This happens e.g. if the user code did not
call pvGetComplete resp. pvPutComplete. In that case we must reset the
request pointer to NULL in order to invalidate the request so it gets
ignored whenever it finally arrives.

The problem was that that once a synchronous request times out, the
variable would never recover from the timeout state. This was because
the semaphores would not be signalled on timeout. However, in order
to correctly discard events (i.e. callbacks from CA) that belong to
timed out requests, we have to remember them. This is what the getReq
and putReq arrays in struct state_set are for.

The fix assumes that CA will eventually call the callback for each
accepted request, even if the channel disconnects. If the assumption is
wrong then we have a memory leak.

This is so I can first start up the IOC, then switch to telnet console,
and then start the tests. The console output (quite verbose for some of
the tests) will be much faster in this way and the tests accordingly
more realistic.

In non-reentrant mode global SNL variables correspond to top-level
static C variables; including their definitions in another C file
makes no sense, since they would be copies, not aliases, for the
variables in the SNL program.

The top-level declaration parts of the generated code are now written to a
header file. These parts are (in the order listed):

- #includes of epicsTypes.h and seqCom.h
- definition of the _seq_var struct
- escaped code segments appearing before the first state set

The stdin/stdout redirection has been removed. Instead code generation now
uses the gen_code function exported from main. Switching between output
files is done using set_gen_h and set_gen_c, which are also exported from
main. This uses global (though static) variables inside main.c for
simplicity.