So, I'm gonna go over a lot of what's already been said, but it's worth taking a look at the background for a lot of these. The most beneficial junctions to grade separate are functionally those with the highest number of conflicting moves per hour.
In the UK, the busiest single junction is actually the one at the south end of Blackfriars station, with up to 18 conflicting moves per hour, but that junction is relatively low speed, and the ATO in the core helps regulate everything so you don't lose any many paths in disruption. Yes, it would be great to grade separate the whole Thameslink Core, but it's not really top of the list.
Another popular one is Working West, which has space to grade separate already but has never been done. Here, there are four conflicting moves per hour, but they have to cross both the down fast and down slow, so they are far more likely to cause conflicts. The junction is also relatively fast, which means being late eats more paths. In fact, it's such a disruption that the whole of the Waterloo main line timetable is written around this one junction.
There are a couple of other important ones that are similar to woking, where there shouldn't be many conflicts, but the conflicts caused are very disruptive, and the junction acts as a constraint on timetabling for a wide area: Uddingston, Newton East, Law Jn and Newbridge in Scotland; Wigston, the north end of Sheffield, Euxton lane, Colwich, Basingstoke East, Wooton Bassett jn and a couple round Birmingham all need doing. These are disruptive if they go wrong, and constrain timetables too much in the 21st century to not be done.
Next there is the second category, big stations that need rebuilt to take in grade separation. The obvious one is Doncaster, but Stafford, Crewe, Stratford, Redhill and Coventry all suffer from this to various extents. These are places where simple grade separation wouldn't work, you'd need to completely rebuild the station in a way that had grade separated routes. At Doncaster for example, there are five possible conflicting routes, with trains in each direction conflicting which is unusual. A grand total of about 17 trains conflicting per hour is getting awfully close to every path conflicting with something else.
Finally there are the nice to haves. These are mainly in plain line and would be amazing if they already existed, but so unbelievably complex to build as to make them virtually impossible. The best example is Leadburn Jn, which I think only has 2tph crossing in each direction per hour, but they have to be timetabled perfectly. If one sneezes on the approach to this junction in the peaks, they'd screw up the timetable for the southern wcml.
Grade separation of a lot of the south London lines comes into this category; how for example could one even begin to imagine grade separation of Lewisham or Sutton, let alone Herne Hill or Tulse Hill? Manchester west end has the same problems.
Personally my vote is for Woking west: it's such a capacity constraint and it's quite likely to go wrong on any given day. Either that or building the almond chord to solve Newbridge and Haymarket East in one, but that's purely due to how much time I spend on trains, waiting for paths through the central belt.
