We've made tons of progress since I joined, redesigning the apps, websites, launching highlights, rebooting our business model, text-to-speech, speed reading, re-writing our parser, re-building our full-text search engine. The list goes on.
We currently don't have an image proxy for Instapaper, so yes when you visit the site we load the original images. We have discussed adding an image proxy but felt it would be a lot of overhead in server costs and maintenance for minimal value.
Additionally, I'm not sure it's fair to represent that fetching images from the original sources that a user saved is tantamount to sharing data with third parties, which has a different set of implications.
I actually liked it that way, that Instapaper stays the same and reliable while other feels the urge to 'innovate' giving it's loyal users only trouble. Scripts that I have written years ago for uploading articles still work and Instapaper is still my Read-It-Later of choice.
As for images - I had to accept that Instapaper works that way - but always had been little annoyed that it is possible to turn off images but that's not persistent option.
That said, speed isn't all that important considering that I do most of my reading on my iOS devices and they have that stuff locally. Still!
But yeah, most likely the code is all over the place.
And the instapaper.com when downloading is sharing data with third parties like there is no tomorrow. Simply it is downloading a lot of crap from original sites (images etc.) which could be used for tracking - no way they could be compliant with GDPR and let user decide which third parties to share data with.