Add ppi argument

[jonmmease]

Closes #85, relies on image-rs/image-png#403 which was released in png 0.17.10.

[jonmmease]

image-rs/image-png#403 was merged, so now we just need to wait for a release of image-png.

Also, adding the pixel dimensions metadata broke the CLI test because they are still doing exact binary comparison. So we need to either regenerate the baseline images, or use DSSIM for approximate comparison like we do with the vl-convert-rs tests.

[jonmmease]

@joelostblom, development wheel are available for this PR at https://github.com/vega/vl-convert/actions/runs/5581857544 (just download and unzip the wheels.zip artifact).

With these wheels, you can add a ppi kwarg to the vlc.vegalite_to_png and vlc.vega_to_png functions. I followed vega/vega-lite#8854 and set things up to increase the scale internally as the ppi increases, so you can leave scale at 1 and increase PPI and the resulting image in PDF export should be the same size but sharper.

When you have a chance, let me know how this works for you!

[joelostblom]

Thanks so much for the quick PR! This is working for me and I am able to adjust both the size and ppi of the charts independently after adding a ppi argument to the the function you created previosuly. As expected, I am getting a constant size at different sharpness in JupyterBooks pdflatex output (to the left in the image below) whilst in JupyterLab, the size of the figure changes instead:

[jonmmease]

Thanks for trying it out @joelostblom, glad it's working as expected! The PR in image-png was merged and the next release should be in a couple of weeks, so I'll put this PR on hold until that point.

[joelostblom]

Just a heads up that I noticed that there was a new release made for image-png last week (or at least there is a new tag at https://github.com/image-rs/image-png/tags).

[jonmmease]

Oh, awesome. I had missed that go by. I'll get this PR updated soon