Passing a type in `shape` kw argument leads to exciting tensors

[samuelpowell]

Passing the type Nothing instead of nothing as part of the specification of a dynamically sized tensor does not error, but produces a tensor of curious dimension:

julia> tf.placeholder(Float32, shape=(Nothing,2,3))
<Tensor placeholder_37:1 shape=(139935212585760, 139935209967216, 139935209967280) dtype=Float32>

Issue requested by @oxinabox following question on Slack.

----------------
Library Versions
----------------
ENV["TF_USE_GPU"] = 1
ENV["LIBTENSORFLOW"] = ~/libtensorflow/libtensorflow.so

tf_version(kind=:backend) = 1.13.1
Trying to evaluate tf_version(kind=:python) but got error: RemoteException(2, CapturedException(UndefVarError(:py_tf), Any[(#11 at TensorFlow.jl:189, 1), (#116 at process_messages.jl:276, 1), (run_work_thunk at process_messages.jl:56, 1), (run_work_thunk at process_messages.jl:65, 1), (#102 at task.jl:259, 1)]))
tf_version(kind=:julia) = 0.11.0

-------------
Python Status
-------------
PyCall.conda = false
Trying to evaluate ENV["PYTHON"] but got error: KeyError("PYTHON")
PyCall.PYTHONHOME = /usr:/usr
String(read(#= /home/spowell/.julia/packages/TensorFlow/q9pY2/src/version.jl:104 =# (Core.:(@cmd))("pip --version"))) = pip 9.0.1 from /usr/lib/python2.7/dist-packages (python 2.7)

String(read(#= /home/spowell/.julia/packages/TensorFlow/q9pY2/src/version.jl:105 =# (Core.:(@cmd))("pip3 --version"))) = pip 9.0.1 from /usr/lib/python3/dist-packages (python 3.6)


------------
Julia Status
------------
Julia Version 1.1.0
Commit 80516ca202 (2019-01-21 21:24 UTC)
Platform Info:
  OS: Linux (x86_64-pc-linux-gnu)
  CPU: Intel(R) Xeon(R) W-2123 CPU @ 3.60GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-6.0.1 (ORCJIT, skylake)

[malmaud]

Thanks for the report!