Tweaks before release.

- Remove pytrace from unit test.  It was polluting the repo.  I saved
  this test case in my scratch repo.
- Comments about release process.

native/fastlex_test.py

@@ -58,40 +58,13 @@ def testOutOfBounds(self):
    print MatchToken(lex_mode_e.OUTER, 'line', 5)

  def testBug(self):
    import os
    pid = os.getpid()
    # This changes the results!  WTF!
    from benchmarks import pytrace
    t = pytrace.Tracer()
    t.Start()

    # woah this makes no sense

    # expected 194, got 197.  So how is expected being corrupted?
    code_str = '-z'
    expected = Id.BoolUnary_z

    # expected 197 got 195 (but it says that is -n, which is 198)
    code_str = '-v'
    expected = Id.BoolUnary_v

    # -n confused with -R
    # 195 198
    code_str = '-n'
    expected = Id.BoolUnary_n

    try:
      tok_type, end_pos = MatchToken(lex_mode_e.DBRACKET, code_str, 0)
      #print tok_type, end_pos
      #print tok_type
      print '---', 'expected', expected.enum_value, 'got', tok_type.enum_value
      #print 'L', id(Id.BoolUnary_z)
      #print 'L', id(Id.BoolUnary_z.enum_value)

      self.assertEqual(expected, tok_type)
      #self.assertEqual(2, end_pos, Id.BoolUnary_z)
    finally:
      t.Stop('bug-%d.pytrace' % pid)
    tok_type, end_pos = MatchToken(lex_mode_e.DBRACKET, code_str, 0)
    print '---', 'expected', expected.enum_value, 'got', tok_type.enum_value

    self.assertEqual(expected, tok_type)


if __name__ == '__main__':

scripts/release.sh

@@ -6,8 +6,11 @@
# Steps:
#   build/doc.sh update-src-versions  (optional)
#   $0 build-and-test  (runs spec tests, etc.)
#     prereq: build/codegen.sh {download,install}-re2c
#   test/wild.sh all
#   benchmarks/auto.sh all on both flanders and lisa
#     prereq: benchmarks/osh-runtime.sh {download,extract}
#   benchmarks/oheap.sh measure
#   benchmarks/report.sh all
#   $0 line-counts
#   $0 build-tree