By dffa203d that reclaim `env` objects from heaps, there's more chance
for `env` objects referenced from fibers may be freed from heap pages.

Fixed a bug in INIT_DISPATCH for non direct threading; fix #4153

Add debug flag to `MRuby::Build`

Keep `tr_pattern` static if possible

Rename `MRuby::Build#libmruby` stuff to avoid confusion

Add missing dependency of source code generator

Suppress sleep test script warning

Sort test script list too

The old condition marks the top-level callinfo even after the
fiber is terminated.

To reduce the env object allocation; ref #4143

The following code mistakenly exits from the function without marking
the env object as `MRB_TT_FREE`.

``` ruby
      if (MRB_ENV_STACK_SHARED_P(e)) {
        /* cannot be freed */
        return; // <- should be `break`
      }
```

fix haconiwa/haconiwa#171

replace quicksort with mergesort.

Refactor .travis.yml to remove duplicated env: MRUBY_CONFIG.

That means entry table should be compacted periodically by `sg_compact()`.