Roads that are designed for higher speed get a design for that: they're straighter, wider, and flatter. That's why higheays, though not flat, are usually less than a 5% grade, why their lanes are wider than city streets, and why things like the Pawtucket S-Curve are so unusual (and why S curves on smaller streets can actually make them safer, by slowing people down). At high speed our vehicles can't accommodate these twists and turns, ups and downs.
But at slower speed, this is not a problem. If the bridges crossed the tracks like a squared off S, there would be no problem. This would work be as a crossing for local traffic: some cars, with a lot of bikes and people on foot.
I tried to diagram it:
ramp off of Harris----------->||
bridge over Amtrak. ||
ramp down to boulevard. ----------->
If you wanted, you could probably do two directional ramps
This tupe of ramp is common with bike bridges, because ADA requirements add length to infrastructure to maintain a mild grade. I see no reason why small, one or two lane car bridges can't be made this way. The land adjacent to the train tracks above Federal Hill has adjacent space because of the demolition of the old fruit depot, and the highway side is even bigger. This would make the most sense as a way to deal with the Dean Street bridge, because although it would be a longer bridge than would be the case if the tracks could be crossed directly, it would still be a lot shorter than if the Dean Street bridge was built from the peak of the hill across.
If we get rid of the faulty idea that the topography and train tracks stand in the way of a boulevard, we also eliminate the need for the halo and for all of the ramps associated with the highway design. We can have a few of these ramp style bridges, at intervals, to carry local traffic each way. Some can be for cars, and some perhaps only for bikes and pedestrians.