I removed the highway=construction tag as it was an erroneous use of the tag. That tag is specifically for roadways that are under construction, to be used with an accompanying construction= tag indicating the final form of the way. It is not to indicate the usage of the way, such as an access road for construction crews to access a site.
In the change set you submitted 5 minutes after posting this question you deleted the way entirely and failed to add the permanent fences you mentioned in your change notes. Inaccessibility is not a good reason to remove a feature which still physically exists. Rather it is more appropriate to separate the impassible portion of the way and add some indication of impassibility such as access=no, preferably with the attending obstruction such as the permanent fence.

Per Slack discussion, platforms on both sides should be able to have the same stop information and be in the same relation without negatively impacting users. Subsequent edit removed multipolygon and restored tags to original platform area, and duplicated tags to additional platform area.

While it's true large parts that I've split here do not have a median, I felt the relatively short distances between median sections and the general trend toward installing more medians along this road warranted committing to fully splitting it. The longest stretch between any manner of median from the Ross Island Bridge to 99th Ave is about 420 meters, a little over 5 Portland blocks. I've done similar work along Airport Way and will review other parts of Powell and a few other areas where there are multiple short sections split into dual carriageways to estimate if they will be better represented as continuous single or dual carriageways. In particular, I aim to clean up areas represented in one of these styles in sections as short as 30 meters.

I took the position of NOAA NGS mark AA1872 that is next to the driveway of the property just south of the intersection of West Younger Avenue and North San Pedro Street, used NOAA's Vertical Datum Transformation tool to convert it's reported coordinates from the NAD83(2011)/NAVD88 datum to the reference frame OSM currently uses, WGS84(G1674)/EGM96, as well as adjusting from epoch 2010.0 when the measurement was performed to epoch 2023.0 to account for the continental drift that has occurred in the interim. I then used the satellite image offset feature in JOSM to align the images carefully with the position of the survey mark. For this project, I used Bing for the more current imagery reflecting recent construction and also that it is coincidentally nearly directly overhead, reducing building shape distortion, and Mapbox for the higher image clarity offered for some features. I also used Google Street View in some areas to supplement some observations where things may be visually vague or obscured. As the NGS mark I used for alignment is not easy to spot in satellite imagery currently available, I used a second mark (AA1873) with a more readily visible position to verify my work using the same coordinate source and transformation process, this one in the gravel lot at the intersection of Wildwood Avenue and Mission College Boulevard. When zoomed in tightly on the current Bing imagery at this location, you may see a roughly circular feature of slightly lighter color than the surrounding material due to the higher reflectivity of the metal cover for the survey mark. In this case, the image was offset by a relatively small amount from stock, about 30 cm. It is common for imagery to have incorrect offsets by as much as a meter, so this sort of fine adjustment can be very helpful. You may also find that city-provided information may vary quite a lot from reality in location, size, shape, and orientation, and that's before we even account for new construction. Some of this is due to continental drift, and some due to low resolution, simplified, or inaccurate data.
Regarding the land use, a majority of the area covered was one large contiguous land use tagged industrial, which was incorrect for most of it. I do intend a second pass over this area to add additional detail including adding back some land use areas with better granularity, such as the parking area for VTA employees at the far end of the rail yard.
As for the parking garage, I overlooked that the garage itself already had a parking amenity tag. As the land use around it would have no other values, I will remove the land use shape.

I'd like to thank you again for drawing my attention to this oversight. Further research showed I was incorrectly using NAD data which, while very high precision, needs conversion to WGS in order to be useful to this project. I will not be reverting these change sets due to the number of changes made to objects to reflect recent construction and added details. Instead, I will be reviewing the areas I worked in previously to correct the NGS mark positions, update the offset database geometry I've uploaded to reflect their actual position (a very small correction from the imagery as presented), and recorrect the positions of objects. When the work is done, their positions will be as precise as I can make them and we will hopefully not have any more obvious mistakes at the edges of the edited areas.

That was probably the one near Mission College, PID AA1873, though I have since also looked at the slightly closer AA1872 near VTA's Downtown Yard. My apologies for the messy boundary in that edit. I usually make the transition on such edits smoother so that they are less jarring.
Regarding the imagery offset references I use, I am currently reviewing the use of NOAA NGS data as presented in their data sheets and how it relates to evolving standards before I continue with this project or make any reversion decisions.
Thank you for your feedback and patience.