{E,E;Δ,Γ} ⊨ ... => {E;Δ,Γ} ⊨ ...
{⊤,⊤;Δ,Γ} ⊨ ... => {Δ,Γ} ⊨ ...

`decide` works much better because of the general lemmas and tacitcs

Add a pre_solve_closed ltac that basically changes the goal from
`Closed X e` to `Closed ∅ e`. It is actually OK in practice.

- Involves bulling the partial bijection algebra from `runST`
- Thanks to Amin for suggestions

From "State-Dependent Represenation Independence" by A. Ahmed, D. Dreyer, A. Rossberg.

Allow to read the location if you have a fractional (non 1)
permission.

Thanks to Robbert