feat: sort pptx shapes to be parsed in top-to-bottom, left-to-right order

[richardye101]

Currently, markitdown parses pptx shapes in Z-order, the order in which shapes are stacked on top of each other starting from the back to the front.

There are repos that parse pptx to markdown which read the shapes in a normal reading order (top-to-bottom, left-to-right order) like https://github.com/ssine/pptx2md/blob/39bef65b312035baeade932aad8d221e37daae5f/pptx2md/parser.py#L249.

There are also stackoverflow posts that explain how to implement this code: https://stackoverflow.com/questions/51999656/how-to-extract-text-from-powerpoint-text-boxes-in-their-order-within-the-presen

I've simply copied over what @ssine has created in his repo, as it's the cleanest implementation.

[richardye101]

@microsoft-github-policy-service agree

[afourney]

It appears that attrgetter is not included.

[afourney]

Thanks! Nice and simple fix.