Nim2 update

[JorySchossau]

I'm updating some old code using this.

Changes:

• replaces finalizer mechanics with Nim 2 destructors.

Blocking:

• Awaiting merging upstream Nim 2 update PR adding destructors to the gmp library Merged! :)

[mratsim]

Thank you!

[mratsim]

Looking more closely, the upstream C bindings should not be overloaded with destructors, those should be added to the Int type here.

See rationale in subsetpark/nim-gmp#2

[JorySchossau]

I agree with you. Adding destructor to Int was actually my first attempt, and the wall I hit was of course since Int is actually a ref and you can't have destructors for refs. So I will go back to my overhaul code to make the better Int and ensure it is used correctly across all the code everywhere it was dereferenced. Suggestions welcome since I may be overthinking it.

[JorySchossau]

I'll have to get back to this later. Here's my progress - there are very few changes. It's probably 99% done - not sure why it's segfaulting and don't have time to dig into it at the moment. I cheated and used a template to bypass modifying any dereferencing code yet, but would be happy to remove it and roll it out across the code later once everything is working.

[JorySchossau]

Okay, that was... educational about the defunct finalizers.

• wrapped the underlying type in an object
• kept the use of ref for consistency and speed
• replaced finalizers with =destroys for the underlying object type
• Bignum Int type now tracks if it is a "view" of a piece of Bignum Rat so the dtor won't release memory the Int does not own (something that finalizers actually solved somewhat nicely)