run-tests: place the .t shell script next to $TESTTMP and with a useful name

--keep can thus keep it around - very convenient for debugging.

tests/run-tests.py

@@ -762,11 +762,11 b' def tsttest(test, wd, options, replaceme'
     addsalt(n + 1, False)
 
     # Write out the script and execute it
-    fd, name = tempfile.mkstemp(suffix='hg-tst')
-    try:
+    name = wd + '.sh'
+    f = open(name, 'w')
         for l in script:
-            os.write(fd, l)
-        os.close(fd)
+        f.write(l)
+    f.close()
 
         cmd = '%s "%s"' % (options.shell, name)
         vlog("# Running", cmd)
@@ -775,8 +775,6 b' def tsttest(test, wd, options, replaceme'
         # similarly, with --debug, output is None
         if exitcode == SKIPPED_STATUS or output is None:
             return exitcode, output
-    finally:
-        os.remove(name)
 
     # Merge the script output back into a unified test