Excel is already programmable via everything from Powershell to VB to F# via its .NET bindings. The problem isn't that anything is missing from Excel, it's that it's used by so many people who are not 10x software engineers that no higher-level approach will ever work with 99,5% of the Excel out there.

The people who are organized enough and use Excel enough to find value in saving formulas in a library are already extreme edge cases...and then you are selling into that tiny market against the idea of using full blown tools like Visual Studio and C#.

And it seems to me that really sound previously snakebit engineers aren't going to go around changing working software that isn't backed up by a suite of regression tests. The problem isn't Excel, it's bad code. The cause of the bad code is bad programming practice, and the reason for that is that the people programming are not programmers and even if they were they're under just give me a kludge time pressure.

Spreadsheets are successful because they're really easy to use. Excel dominates because making one that's userful for important work is really really hard.

Good luck.