Type inference using type annotations

[rahul-kamat]

Performs type inference using type annotations.
A dictionary with variable names mapped to a list of possible types is added to the AST nodes.

Example

def f(x: tf.Tensor, a: int):
  if y > 0:
    y = x + y
  return y

The AST will be annotated with:
y = x + y:

___pyct_anno={SCOPE: Scope{r=(x, y,), w=(y,)}, 'type_anno_read': {x: {tf.Tensor}, y: {int}}, 'type_anno_write': {y: {tf.Tensor}}}

return y:

___pyct_anno={SCOPE: Scope{r=(y,), w=()}, 'type_anno_read': {y: {tf.Tensor, int}}, 'type_anno_write': {}}

Note: This is experimental and does not infer types for variables assigned to a python object.

[rthadur]

@rahul-kamat can you please check ubuntu sanity build failures ?

[mdanatg]

The PR is still valid, but we've been delayed on the integration.

[gbaned]

Hi @rahul-kamat, @mdanatg I'm going to go ahead and close this PR, because it seems to have stalled. If you're still interested in pursing this (and responding to my comments), please feel free to reopen! Thank you for your contribution!