Clarification: Having read osm.org/note/1791672, I confirmed that the sign said "Elementary" and not "High" school, but I didn't check the spelling of the rest of the name.

FWIW, I was mistaken: OsmAnd seems to recognize access:conditional. I think this code (https://github.com/osmandapp/OsmAnd/blob/feadaa0da5804d9f3312e6df852ef96093fb82ec/OsmAnd-java/src/main/java/net/osmand/binary/BinaryMapRouteReaderAdapter.java#L237) is involved, though I haven't traced the entire data flow.

Apparently access:conditional is well-established, but I can't find a router that supports it to test this change. :(

The segment of eastbound O Street that I changed is somewhat artificial to OSM's representation of the complex intersection, and I didn't see any turn:lanes signage for it specifically. But the combination of the turn:lanes I added with the existing value for southbound 1st Street seems to correctly model the reachability of eastbound New York Avenue and O Street from the left three southbound lanes of 1st Street (see osm.wiki/Relation:connectivity#Basic_summary). Indeed, it fixes the lane guidance in OsmAnd for the maneuver from southbound 1st Street to eastbound New York Avenue. It might be better to use connectivity relations directly from southbound 1st Street to eastbound New York Avenue and O Street, but fewer routers support those, and I'm unsure of the exact target lanes on eastbound New York Avenue and O Street, which would need to be stated in the connectivity relations.

turn:lanes:conditional seems to be very unofficial (in particular, I'm unsure how I'm supposed to protect the semicolon in the conditional value; I took a look at taginfo for prior art), but that's the fun of OpenStreetMap, right? I'm having fun anyway. If enough instances of turn:lanes:conditional are added, maybe routers will be motivated to learn to parse it.

Concerns:
- In several cases, I entered turn:lanes with a number of lanes that does not match the "lanes" tag because additional lanes for turns begin shortly before the intersection, but I didn't know the exact place where they begin in order to split the way and update the "lanes" tag. I hope this is OK: I think most consumers of the data will do something reasonable.
- Northbound Connecticut Avenue at Jones Bridge Road: I think it only makes sense for the rightmost lane to be "through;slight_right;right", but I couldn't confirm this by signage.

Information about the construction project is at https://www.montgomerycountymd.gov/whiteflint/ . I used the vehicle detour map there to plan where to survey, but I didn't copy any information directly from that map because I was unsure of the licensing. I used GPS traces to locate the closures and tried to get the connectivity right, at least for motor vehicles.

Review requested!!! I've probably done some things not according to convention, but I hope this is at least much better than the status quo. And someone who cares about pedestrian and bicycle access is invited to update that part. Discussion of the complete update to this area beyond this initial changeset can take place at osm.org/note/2590993 .

Signs I saw:
- Entering the changed segment from the traffic circle: "One way" with a forward arrow.
- Attempting to turn east onto the segment from its west end or from the way between Kenamond Hall and Dining Hall: "Do not enter".
I'm not certain the entire segment is intended to be one-way, but I thought it was likely enough that marking it one-way was better than adding many turn restrictions given the lack of support for "no_entry" in routers (https://help.openstreetmap.org/questions/49513/should-i-use-the-no_entry-turning-restriction).

- The time-based restriction at westbound Blackthorn St is "do not enter" from any direction (surveyed 2021-02-25).
- AFAICT, there isn't an unconditional "no U turn" restriction from southbound Connecticut Ave at Dunlop St (surveyed 2021-02-24). The time-based "no left or U turn" at this intersection is correctly coded in relation #12252810. There is an unconditional "no U turn" at the fire station entrance just north, but that cross street isn't on OpenStreetMap yet at all.

I don't think this change is particularly important: most of the minor roads don't have turn:lanes tags. I'm just having fun because the local transportation department recently signed these turn lanes at my request.

I know by survey that a left turn from northbound Citadel Avenue to westbound Old Georgetown Road is blocked by the median on Old Georgetown Road. I don't know how far the median extends; I've taken a guess from the Bing imagery. I wasn't at all confident of how to make this edit, but I hope what I did will be better than the status quo until someone more experienced takes a look.

Oops, the changeset description is backwards: the segment was moved from Greenhill Road to Old Lancaster Road.