[fdupress]

Please feel free to ignore (and close) this if a better upgrade is possible.
In particular, why3 1.9 seems to bring new features that could be of use if we choose to adopt them...

[fdupress]

The result from running nix flake update is a lock file pointing to a 6-month old commit in opam-repository, unlikely to have the why3 release from last week. Is the right way to go about this to simply rebuild and relock from scratch? I don't know.

[fdupress]

OK. The flake is fixed.

This PR could possibly be split into one that fixes the flake (by pinning a version of opam-repository that makes sense) and possibly also updates the stable (not done here) and alt-ergo (which I should have done when I bumped it), and then one that bumps why3.

I'm enclined to not bother.

[strub]

I amended the commit with two small changes: flake.nix now asks for nixpkgs/26.05 for stable, which is what flake.lock already records (the lock and the flake disagreed), and the %r special case passes ~is_neg:false with an assertion, since f_int never builds a negative Fint (negative integers are [-] n), and why3 1.9 rejects a negative mantissa anyway.