…1.8.5 release.

Lots yet to do.

…rsion of spidermonkey can be used easily.

- Added new calling sequences with jsid instead of jsval
- Temporarily removed checks for JS_THREADSAFE, but a lot has changed in 1.8.5 that requires review
- Changed malloc() to calloc() to assure that undocumented structure members are initialized to zero.
- Eliminated internal "js_" calls and replaced with new "JS_" API equivalents.

…yobject.c)

- Deprecated old embedded libjs build and removed old libjs code.
- Added ability to add finalize functions to general hashobj wrappers to handle synchronization of tasks between js and python
- make_object now NEVER returns an object which is not of the requested type.   There was erroneous code which unwrapped existing py objects which were causing memory overwrites with certain types of return values.
- Added unwrap_pyobject() to handle cases where we want to unwrap rather than double-wrap python objects passed back from js
- Still some bugs in iterators.
- Still some problems with old memory and runtime checks do to the Operation Callbacks change in 1.8.5.

- Fixed bug in iterators which referred to the wrong
  object in the new JSNative calling sequence.

  segfault randomly.
- Fixed jsval references to use macros properly. It's likely there were
  some bugs hiding since the comparisons made using previous API techniques
  were no longer valid.

…ycles internally

  when wrapped objects get assigned to global storage.  This means it is the responsibility
  of the owner of the context to make sure the global object is not deleted, but in the case it is
  the library will simply failsafe by no longer supporting globals for access or modification.
- Changed StopIteration exception to use 1.8.5 library call.
- Fixed reference counting problems with iterators.

Otherwise, the global object is kept as a reference and may cause leaks
if populated with wrapped python objects.

…k routine.

Now, you can either return "False" or raise an exception to indicate an access check failure.