Add handling for non-NumPy types in Shape/Shape_i

This change allows `Shape`/`Shape_i` to handle non-standard `Op` values, like
the `None` and `list` values produced by `NoneType` and `MakeList` `Op`s.

The one drawback to this change is that `theano.scan`, i.e. the `scan` function
exposed by `theano/__init__.py`, shadows the `theano.scan` sub-package.

In other words, modules in the `theano.scan` sub-package cannot be accessed via
the top-level `theano` module object (e.g. `import theano; theano.scan.op.Scan`
won't work).

This is a minor inconvenience, and, since internal library code is generally
expected to import objects directly from the modules in which they're defined,
the appearance of this problem will serve as a welcome warning.

This replaces `theano.tensor.basic.dot` with the old `theano.dot`, removes the latter, and renames `theano.tensor.basic.dot` to `theano.tensor.basic.dense_dot`.

Remove theano.compat and other Python 2.x compatibility code

Co-authored-by: Brandon T. Willard <971601+brandonwillard@users.noreply.github.com>

Refactor tests.tensor.test_basic

- Remove unused tests and unnecessary helper functions
- Rename camel-case tests
- Use pytest.raises

- Use `isort` to sort imports
- Remove import cycle in `theano.gpuarray`
- Remove some Python 2.x code using `pyupgrade`.

If we use Python varargs, we're limited in the number of arrays/shapes that can
be processed/given.  Now, there's a `broadcast_shape_iter` that takes a
collection of arrays/shapes and doesn't face this restriction.