partial merge and full RCS ID sync with oksh

merged:
• new regression tests
• check.pl (tests/th) better tmpfile handling
• exec.c 1.50: POSIX specifies that for an AND/OR list,
  only the last command's exit status matters for "set -e"
• ksh.1 1.147: document the above
• eval.c 1.39: “Make $(< /nonexistent) have the same behaviour
  as $(cat /nonexistent) wrt. errors (do not unwind and do not
  treat this as fatal if set -e is used).”
  ‣ additionally make shf_open() return errno and actually show
    the error message from the system
• regression-39 test: remove the “maybe” marker
  ‣ but decide on correct POSIX behaviour

already been fixed in mksh:
• check.pl (tests/th) exit 1 if tests fail

not merged:
• main.c 1.52: run traps in unwind() before exiting;
  I’m pretty sure this is already working as-should in mksh
• eval.c 1.38: “Commands executed via `foo` or $( bar ) should
  not inherit "set -e" status.” As discussed in IRC, this is
  just plainly WRONG.

shf.c

@@ -25,7 +25,7 @@
 
 #include "sh.h"
 
-__RCSID("$MirOS: src/bin/mksh/shf.c,v 1.60 2013/06/01 20:34:01 tg Exp $");
+__RCSID("$MirOS: src/bin/mksh/shf.c,v 1.61 2013/07/21 18:36:03 tg Exp $");
 
 /* flags to shf_emptybuf() */
 #define EB_READSW	0x01	/* about to switch to reading */
@@ -52,7 +52,7 @@ shf_open(const char *name, int oflags, int mode, int sflags)
 	ssize_t bsize =
 	    /* at most 512 */
 	    sflags & SHF_UNBUF ? (sflags & SHF_RD ? 1 : 0) : SHF_BSIZE;
-	int fd;
+	int fd, eno;
 
 	/* Done before open so if alloca fails, fd won't be lost. */
 	shf = alloc(sizeof(struct shf) + bsize, ATEMP);
@@ -64,16 +64,20 @@ shf_open(const char *name, int oflags, int mode, int sflags)
 
 	fd = open(name, oflags, mode);
 	if (fd < 0) {
+		eno = errno;
 		afree(shf, shf->areap);
+		errno = eno;
 		return (NULL);
 	}
 	if ((sflags & SHF_MAPHI) && fd < FDBASE) {
 		int nfd;
 
 		nfd = fcntl(fd, F_DUPFD, FDBASE);
+		eno = errno;
 		close(fd);
 		if (nfd < 0) {
 			afree(shf, shf->areap);
+			errno = eno;
 			return (NULL);
 		}
 		fd = nfd;