however at higher bitrates and resolutions these aretifacts should be too small to be easily perceptable.
Agreed and this is key to the overall look of the video. It is always important to use your best codecs if quality is important and then those artifacts that cause left - right differences will not affect overall visual in 3D.
I once experimented with an example of left right image differences that were major. Here I did two tests. You may wish to experiment in a similar way.
I intentionally took a right image and modified it's color , contrast, and brightness then paired them and observed for what difference was in 3D.
The color did cause a bit of problem as colors blended to a new hue except where the images diverged and here the ghosting of double image outlines surfaced. In contrast and brightness differences, the same ghosting happened but the objects all just looked additive in contrast and brightness. e.g. a pitch black became a dull gray. Note in my test I made obvious radical changes that were well outside the normal artifact range.
Where this issue did become critical had to do with the left right image differences that were actually captured during recording. For example, I tested with glare elimination using circular polarizing filter. In this case a different level of filtration was recorded for left and right lens because the filter was in a different setting due to each camera lens being off center to the single filter rotation on each lens. Glare was never completely eliminated but worse the glare was recorded with different depth than the object generating the glare. Like a windshield, the glare was no longer on the glass but either under it or above it. This was weird and I found no easy solution so I chose to avoid using any filter that could not function equally for each off center lens. UV and ND filters work great, Gradient filters can work in horizontal rotation alignment but not if rotated at an angle. Polarizing filters don't work at all for 3D where one filter is used for both lenses. It works fine when each lens has it's own filter adjusted individually.
Finally, anytime any object is present in one image and not the other they will combine when paired and the result will display in 2D at the screen plane. This can be a problem, or not with occlusion in the paired images.