On my phone so will try to be brief. The first big question is how much time and resources you want to put into yeast management. It’s not inconsiderable and one reason why so many small breweries use dry yeast.
Second- attenuation and dropping are generally mutually exclusive, which is why British breweries historically used a multi strain that included yeast that were good for one or other. Multi strains are a pain to manage in conical which is why Whitbread B (parent of wlp007 and 1098; 1099 is a completely different Whitbread strain) became so popular during the industrialisation of brewing in the 1960s as unusually it did both fairly well.
Third - there’s lots of other considerations beyond just taste - when Burton Bridge set up in the early 1980s they set an initial filter of the NCYC based on eg chain forming for easy identification under the microscope and only then did a taste off. Their situation was different as they were aiming for the cask market but still worth a read by anyone planning to open a brewery
http://www.burtonbridgebrewery.co.uk/History/Index.shtml
By accident they ended up with a killer yeast which was a bonus- I’m trying to get hold of some as it sounds a fun yeast.
Anyway, combine those first two and think about blends of dry yeast - I’d definitely put 90% S-04 / 10% T-58 on the list, which seems to be the core of the Treehouse blend (which may also have a pinch of WB-06 as well, but I’d KISS to start with). Gives you a lot of flexibility to just use one or other in appropriate styles - supposedly de Struise use T-58 in Pannepot which is a world-class Belgian dark.
Two other slightly left-field choices which are on my list to try in NEIPAs are the Rochefort yeast (wlp540 etc - which is POF- and a relative of Ringwood/East Midlands so not “Belgian”) and Mangrove Jack M15, although the latter might be a bit of a pain in a brewery environment not least because it only tolerates 8% alcohol.
As an aside - although Fullers claimed to select one strain from their multi strain when they moved to conicals in the late 1970s, the genetic evidence suggests wlp002 split off from Whitbread B around that time... Although S-04 has generally been linked to Whitbread B (ie 1098-ish), recent genetic studies, which may not be entirely reliable, suggest that it too has come out of the Ringwood and Wlp039 family.