* Implement support for Some.Component style JSX blocks

* Add failing test for string interpolation

* Add failing test for js functions as props

* Add failing test for js functions as props with returns

* Make value parsing for JSX props more flexible

* Don't lint fixtures file

* Add nested object props fixture

* Add another fixture for randomly placed brackets

* Add link shortcut to fixture

* Remove object props since it is a subset of js props

* Update packages/remark-mdx/test/fixtures/inline-parsing.js

Co-Authored-By: johno <johnotander@gmail.com>

* Add publish script and configuration for ci

* Add travis config to skip cleanup

* Add .npmrc file for publishing through travis

* Get correct tags before publishing with lerna

* Improve import and export parsing with babel

This introduces babel for better handling import and
export parsing. Anything that is an import/export block
will be parsed by babel and split up ensuring that
edge cases will no longer occur based on how the
JS is grouped. It also makes parsing the export default
nodes more robust.

- #345
- #340

* Add multiline fixtures

* Preserve positional information for AST nodes

In order to maintain positional metadata for AST nodes
that are parsed, we need to ensure that each import/export
is eaten in order as a correct subvalue.

Since the remark eat function requires the string to be
consumed in order, with no missing characters, the babel
plugin now returns the start value which is used to sort
the nodes and partition the input string. They're then
combined again and returned as nodes that can be eaten
in order by the tokenizer.

This also comes with the benefit of no longer changing the
input code which occurs with the prior babel generator.

* Remove unneeded babel generator dep

* Simplify partition function

* Make naming consistent

* Add snapshot test to verify positional info