we don't need it anymore now that we eliminated nearly all pointers to nodes.

move constants into CodeConstants, call BoxedCode destructor, cleanup BST nodes

A number of small fixes from the dropbox testsuite

This warning is quite annoying right now

- Fix a test to do the correct number of iterations
- Catch more exceptions in PyDict_GetItem/createModule (now that more things can throw)

BST: convert all nodes to directly operate at vregs instead of names

**basic design:**
This PR changes our BST nodes to directly operate on vregs instead of pointers to other nodes and names (except a few exceptions: `BST_Invoke`, `BST_MakeFunction` and `BST_MakeClass` which still needs to get converted).
Most nodes got a destination vreg and one or more source vregs. Currently all of them are 32bit long but I plan to store them more compact very soon. Some nodes support a variable size of operands (e.g. the tuple node) but the size can't change after creating the node. I removed several unneeded opcodes and split a lot of nodes into separate opcodes (it may make sense to split them even further in the future).
Generally all instructions except `CopyVReg` kill the source operand vregs except if the source is a ref to a constant. If one needs the preserve the source vreg on needs to create a new temporary using the `CopyVReg` opcode.

There is a special vreg number: `VREG_UNDEFINED = std::numeric_limits<int>::min()`.
- when it's set as an operand vreg: it means that this is a not-set optional argument. (e.g. for a slice which only has `lower` set, `upper` would be `VREG_UNDEFINED`)
- if it's the destination it's means the result value should get immediately killed (e.g. `invoke 15 16: %undef = %11(%14)` this is a call whose result gets ignored)

all other negative vreg numbers are indices into a constant table (after adding 1 and making them positive).
(e.g. `(4, 2, 'lala')` generates:  `%undef = (%-1|4|, %-2|2|, %-3|'lala'|)` this creates a tuple whose elements are the constant idx -1, -2 and -3. In order to make it easier for a human to understand we print the actual value of the constant between | characters)
- constants can be all str and numeric types and 'None'.
- every constant will only get stored once in the table

this reduces the total memory usage by about 20% currently but I'm very sure with the future changes it will be significantly lower.

**near future:**
- change the jump and branch instruction to reference `CFGBlocks` by index.
- store all `InternedString` inside a table and use indices into the the table to access them.
- remove the 'BoxedCode*' member
- devirtualize the classes
= with this changes the bytecode can get freely copied around (only need to update the CFGBlock table) which allows us to attach the directly next to each other.

- I plan to use one bit of the the opcode to mark the instruction as only requiring 8bit vreg operands (which should handle the majority of cases with 128 temps and 127 constants + 1undef vreg value)
- another bit will get used to specify if this instruction is inside an `invoke`. if this bit is set there are 2 one 1 or 4 bytes long block indices directly behind the instruction.

- serialize the bytecode to disk. (maybe serialize the constants using pickle)

**thing which need to get improved**
- currently the constant table get's attached to the `BoxedModule` maybe there is a better location, I also needed to pass the `BoxedModule` into some functions e.g. BST printing because otherwise we could not pretty-print the constants
- `BST_Name` is not an opcode it's just used to initialize the arguments when a function get's called and stores where and how the arguments need to get stored.
- more consistent opcode names and rename `TmpValue` to something better
- we currently don't print the `InternedString` name - we only print the vreg number

**additional changed made which are hidden in the large diff** 👎
- removed unused code initializing the items of `BST_Dict` (we use/used separate  assignments to add the items)
- lower `ExtSlice` inside the CFG phase to a tuple of slices
- separated opcode for load subscript when it needs to be a slice and when it's only lower and upper (=`__getslice__`) before this got handled in the interpreter/jit
- generate a constant `None` load inside the CFG when `None` gets loaded by name

Specifically, when passed to PyArg_ParseTuple when a dict argument
is asked for.

google's protobuf library gives away one too many refs to one of its types.

This workaround feels excessively specific, but this is a common mistake to
make, and it works just fine in CPython, so let people get away with it
as well in Pyston.

We weren't properly reducing the stack depth when yielding from a generator,
meaning that the recursion depth was effectively decreased by the number
of active generators.

Too many things want this to be able to remove it completely.  Surprisingly,
having imp.get_magic exist but return a changing value seems to make all those
libraries happy.

A function at risk of "naive misuse", it is only accessible via the C API.

This commit adds basic support for it using the same mechanism we use for
signals.  We also have the GIL-check mechanism, but that would be a bit more
work to get working right now due to the fact that our GIL-checks don't support
throwing exceptions.  Doing the async-exc check during signal checking means
that we will throw the async exc faster than CPython does.  It also means
that there are some pathological cases where with a lot of threads and a lot
of async excs we will probably have much worse performance.  But as long as they
are rare I think this commit shouldn't add any steady-state performance costs.

Specifically, for the case that locals==NULL (which Cython exercises), which
wasn't previously working.

now using ctypes to test the c api

We fast-pathed "type has __init__ but not __new__" by saying
that the default __new__ would always succeed, but this isn't true for abstract classes.

Add test for typing.py and fix the issues it uncovers

Otherwise they would inherit object's __doc__

Update lz4 to a version compatible with gcc 4.9+

Not 100% sure of the reason but lz4 gets mis-optimized (or "has UB"
depending on who you ask) in the version that we're using, so update
to a version that has a fix.

Make these properly CAPI functions

Change section memory manager to fix performance problem in long-runn…

remove RODataMem.FreeMem.clear(); and replace return true to RELEASE_ASSERT(0, "finalizeMemory failed")

Add type.__instancecheck__ and __subclasscheck__

We supported classes that overrode them, but didn't provide a
default implementation in case people called it by hand.

The tricky part is making sure that these new additions don't
mess with our "does this class override __instancecheck__" optimizations.

Fix docs about stats option

Take a pass over the CPython tests