- revised destroyhandler to be more efficient, in view of how frequently
  env.destroyers() was being called.
- fixed errors in tensor.flatten(), but also changed semantics of the optional
  shape parameter.
- added testcases of the previous merge optimizer fix.

-started writing missing test.

was being copied, but not the value itself

bugfix. If you used Value, TensorValue, tensor.value(), the value was not considered as an input. So this was making the inplace_opt optimizer optimize it while it should not be doing it.

- added allow_gc option to PerformLinker and OpWiseCLinker, which results in
  temporary results being freed as soon as possible.  This can have a negative
  impact on performance (we have to re-allocate outputs) but allows us to
  evaluate graphs piecewise, even if not all temporaries fit in memory at
  once.

- modified Function so that internal references to `required` input arguments
  and computed output arguments are removed after a call.  This means that
  these values are not stored by pickle.

- modified MergeOptimizer to use a more efficient algorithm.  The previous one
  was at best quadratic in depth, and at worst exponential in depth.  The new
  one should be worst-case linear.

* Implemented TensorDot op (see numpy documentation for details)
* Added respective test cases under tests/test_basic.py
* Print node now accepts a list of things to print, which can be input node
attributes or functions (eg. "__repr__"). By default, will print input.__str__() as
before

* ValueError: ('invalid number of arguments', mul(Subtensor{0}.0, 18))
This error happened after serializing a model. Scalar ops taking a variable
number of arguments had their ufunc set to that op with -1 input parameters, and
nout output params. Code has been added to check this condition and set ufunc to
None which postpones the building of the ufunc to when we know the actual number of
params (or something like that -see OlivierB.).

* Optimization bug which seemed rather serious.
Fixed bug with Canonizer and DimShuffle
Symptom: ValueError: ('Input dimensions do not match.'