Pipe changes throughout suggested by Eric Fifer <EFifer@sanwaint.com>

* debug.cc (threadname_init): Pass name of lock as arg 2 of new_muto.
* malloc.cc (malloc_init): Ditto.
* sigproc.cc (sigproc_init): Ditto.
* exceptions.cc (events_init): Ditto.
(call_handler): Eliminate special case for hExeced.  Report locked thread in
debugging output.
* fhandler.cc (fhandker_pipe::fhandler_pipe): Propagate device type to base
class.
* fhandler.h (fhandler_pipe): Ditto.
* hinfo.cc (hinfo::build_fhandler): Pass specific type of pipe to constructor.
* spawn.cc (spawn_guts): Eliminate dependency on signal when waiting for
subprocess.
* strace.cc: Remove obsolete #ifdef.
* sync.cc (muto::muto): Save the name of the muto.
(muto:~muto): Also release the muto.
* sync.h: Add a muto name field.
* select.cc (peek_pipe): Avoid doing a PeekNamedPipe on the write end of a
pipe.

winsup/cygwin/fhandler.h

@@ -327,7 +327,7 @@ public:
 class fhandler_pipe: public fhandler_base
 {
 public:
-  fhandler_pipe (const char *name = 0);
+  fhandler_pipe (const char *name = 0, DWORD devtype = FH_PIPE);
   off_t lseek (off_t offset, int whence);
   /* This strange test is due to the fact that we can't rely on
      Windows shells to "do the right thing" with pipes.  Apparently