[TensorFlow] Support for Shape and Reshape in TensorFlow frontend

[ghost]

Thanks for contributing to TVM! Please refer to guideline https://docs.tvm.ai/contribute/ for useful information and tips. After the pull request is submitted, please request code reviews from others in the community.

[ghost]

@tqchen @yzhliu @srkreddy1238 Could you please suggest the reviewers for this PR? I have taken care of comments from Siva.

[ghost]

Please also have a look at the fix done in #1894 related to usage of shape dict only when the shape operator is converted into NNVM symbol inside tensorflow.py

[tqchen]

please rebase against the current master @wweic @srkreddy1238 please review

[srkreddy1238]

Reshape require further handling to support InceptionV4 & ResnetV2.
#1923 handles it as required and is verified too.

[wweic]

I plan to look into #1864, this PR seems to be in the same area. Could you share what testing have you done for this PR? The reason I ask is because I'm not able to pass all unit tests in nnvm/tests/python/frontend/tensorflow/test_forward.py(even in master branch!). We should fix the failure if that's the case and enable tests under nnvm/tests/python/frontend/* in CI.

[ghost]

@wweic Please look #1923 as pointed by Siva above. They are comprehensive changes and superset of this PR. I will close this PR (#1884 ).