I disagree on Zibanejad. He has 5-years remaining on his deal after this season at $8.5M per, you cannot have a dead cap hit of $4.25M per for that long, that would be a very poor decision, and quite frankly awful asset management driven by emotion, which is a terrible way to operate as a leader.
My take on Zibanejad is the Pres/GM and the coach (whomever they will be - Drury/Lav or other people in those roles) need to have a deep discussion with him after the season and tell him bluntly this season was not close to good enough. He needs to work harder in the off-season and during the season, he needs to shoot the puck WAY WAY WAY more than he does, and he needs to bring a higher level of energy/passion/compete moving forward. If he doesn't do those things, his playing time will suffer greatly moving forward. You hope those words, and holding him ACCOUNTABLE, will sink in and he rededicates himself to being a top player next season. I think this is the best option, unless by some major miracle you can get him to agree to be traded, and you find a team willing to take his whole cap hit.
As for Kreider, you have more flexibility, I think you can trade him much easier with 2-years remaining on his deal after this season at $6.5M per. Even with a 15-team no trade list, I'd expect him to be movable without having to retain money, or at most maybe 25%. The GM can get this done I believe.
As for the draft pick, be careful what you wish for. By hoping the Rangers keep their pick this season, pick # 11 or 12 or 13, you are essentially saying they will be much better next season. I think they can/will be, but only if they are bold and aggressive this summer. The worst thing would be the Rangers keep pick # 12 this draft, and watch the Pens get pick # 4 from the Rangers in next year's draft, that would be a disaster. So, IF the pick stays with the Rangers, it should be an absolute major driver for Drury (or whomever the GM is) to ensure the team is very good next season and he needs to be aggressive this summer to make it happen.