Interesting, but holy moly, that looks like a lot of work for minimal benefit. Alas, too much dynamic everything.
To consider one example, converting len(“123”) to 3, that’s something I might do with barely more than a sed script. There’s a mention of some previous efforts, but I’m not sure what they did. The possibility that somebody replaces the len() function (what!) would mean you can’t always apply a simple text substitution, but how frequent is that? If you have the kind of code that’s performance bound by static len() calls, would you consider using a simpler code rewriter instead?
We already have a Python interpreter that’s 5× faster than the standard one: PyPy. Apparently making Python developers waste years of effort porting to a backwards incompatible Python 3 was no big deal, but switching to a fundamentally superior interpreter that breaks a much smaller codebase (C extensions) is a no go for python-dev team.