fix number of purity args on 1.11+

[KristofferC]

Fixes #114.

[topolarity]

The new effects need to be tightened (as suggested) before this will be foldable, but LGTM 👍

Thanks!

[topolarity]

Thanks @chriselrod !

[chriselrod]

No problem!

The auto-merge won't work while tests are failing.

There were unfortunately some regressions that I never got around to looking at.
I'd suggest either

1. Override to ignore branch protection and merge anyway.
2. Fix those tests...
3. Turn off the tests?

As of Julia 1.12, the tools will exist to drop RecursiveFactorization (and thus the JuliaSIMD ecosystem) from the SciML ecosystem by open sourcing a certain package (that requires LLVM17 for replacing RF), so folks will have to consider cost/benefit in the approach you take for getting important packages running on Julia >= 1.12.
At the very least, I'd consider the possibility of doing that, to encourage users of JuliaSIMD to pay for support if they want support, rather than take it for granted.
If it isn't important enough for anyone to be willing to pay for it, why bother spending time on it?