Replace intarray with gc_vector<int>, combine varpower types + range-based loops

[mahrud]

This commit replaces the type intarray throughout the engine with a specialization of std::vector which uses the GC allocator:

• replaced intarray with gc_vector

Note: gc_vector<T> is basically VECTOR(T), so at some point I'll also replace all VECTOR(T) with the new gc_vector<T> syntax.

This commit replaces the varpower types into a templated type that uses gc_vector<int>:

• combined varpower types into ExponentList

The last two commits are pretty minor changes:

• switched to range-based loops for Nterms
• renamed and moved IM2_kernel_of_GB -> rawKernelOfGB

[mikestillman]

For the replacement name for varpower monomials, how about SparseExponents or ExponentMap? The main problem with this last name convention is that in the noncommutative case, it is not a map at all... I guess ExponentList is perhaps more correct, but doesn't necessarily convey any other information (e.g. that it is only representing those exponents which are non-zero).

[mikestillman]

@mahrud I'm going to start reviewing this so we will be ready once we choose the replacement name for varpower monomials

[mahrud]

@mahrud I'm going to start reviewing this so we will be ready once we choose the replacement name for varpower monomials

It might be easier to accept #2769 first, which currently contributes to a big chunk of the diff in this PR.

[mikestillman]

@mahrud Thanks for the heads up, I mistook this one for the next one in the series of pull requests. I'll review #2769 first.

[mikestillman]

@mahrud It looks like there are many conflicts now that I rebased and merged #2769 . After you take a look at these, I'll start reviewing this one.

[mahrud]

@mahrud It looks like there are many conflicts now that I rebased and merged #2769 . After you take a look at these, I'll start reviewing this one.

Done!

[mikestillman]

Hi @mahrud, thanks for making these changes! I have flagged a few places where there might be an issue. Once you look at those, I'll accept this pull request. I'd also like @antonleykin to see how his NAG code had changed, so it won't come as a surprise later!