On the simple test of `pyston -c "import pip"`, it reduces memory
usage from 194MB to 153MB (20% of the memory previously was cached
analysis data).  It also increases our benchmark geomean by ~1%,
which isn't great, but I think we can get that back eventually and
I don't think it's worth blocking this memory improvement for that.

I guess in CPython, __file__ will be the path to the pyc if that's how
the library was loaded.  (We don't do that but that difference isn't the
issue here.)  So that makes these tests, which print __file__, sensitive
to whether the last run of the test was through cpython (pyc exists and
__file__ ends in pyc) or through pyston (pyc exists but was wasn't valid
for cpython, and __file__ ends in py)

Just do some canonicalization when printing

- make heap dumping work again
- refactor some of the heap statistics to work in OOM scenarios
  (ie be able to not collect the ones that would require allocation to track)
- collect some hcls statistics as well
- print stats to stderr rather than stdout
  - I think this makes more sense since it is not really part of the program
    output; it also makes it more likely that "PYSTON_RUN_ARGS=s" will work

prohibit _PyTuple_Resize for objects that are subclasses of tuple

implement `real` and `imag` for `int`, `long`, and `float`

ignore untracked files in the gflags submodule

remove ununsed class BoxedUnicode

Add PyGC_AddRoot calls any time we assign to a static variable

It seems like there's only one potentially bad one (zip_directory_cache),
and the rest are more or less benign (most of the occurrences are added
to the module as well as attributes), but it's probably good practice.

run gflags unittests

gflags test is passing

Also add file.name

It provides overrides for some of the values
in the "string" module, which ends up mattering
for the gflags tests.

Just another HAVE_foo define we were missing.

Otherwise git keeps asking me what I want to do with the modifications
in the pycrypto submodule.

Conflicts:
	src/runtime/types.cpp

Add a pycrypto integration test

A quick (40s) sanity test to run before sending it off to travis-ci
for the full test.

Fix misc smaller issues

This lets us support `exec compile(ast.parse(s))`.

Also, add 3-arg eval() support, and simple support for the _ast module.

gflags support

Put these defines earlier.
I don't like that Make doesn't warn on undefined variables

With the goal of making `assert obj.__dict__ is obj.__dict__` work.
I haven't seen people do that, but people (gflags, doctest? others?)
definitely need `assert globals() is globals()` to pass, which uses
the same mechanism.

The CMake build system doesn't have support for building the extension
modules, so use the self-hosting ability here.

Also, remove ext_pyston from run_deps since I think this would
become pretty anoying for debugging, where you can't run your test
if setup.py wouldn't run.  It's pretty rare that we actually use these
modules anyway.

LLVM has a limit of about ~255 operands per machine instructions, which we could exceed
with our patchpoints usage.

and add a pycrypto integration test.

In additon fix an assert inside CompiledFunction::speculationFailed

add pkg-config to the install instructions

pyc number bump since we fixed the tuple parsing issue