Update FastDEC after ACSet updates

[GeorgeR227]

This is meant to replace views in the FastDEC module with regular ACSet accessing. I've also taken the liberty of cleaning up some strange logic.

[GeorgeR227]

A few benchmarks of note. Overall performance otherwise is preserved.

Old (09ec35f):

Operator: Geometric Hodge
Variant: New Form-1, [0.935041 ms, 2.212816 MB]
Variant: Old Form-1, [53.808375 ms, 54.740096 MB]

Operator: Wedge Product
Variant: New Form-0, Form-1, [0.052583 ms, 0.308864 MB]
Variant: New Form-0, Form-2, [0.339291 ms, 0.415056 MB]
Variant: New Form-1, Form-1, [0.1625 ms, 0.35056 MB]
Variant: Old Form-0, Form-1, [8.164167 ms, 8.05008 MB]
Variant: Old Form-0, Form-2, [46.512188 ms, 83.6008 MB]
Variant: Old Form-1, Form-1, [50.183625 ms, 56.1576 MB]

New (67ad092):

Operator: Geometric Hodge
Variant: New Form-1, [0.9962295 ms, 1.8428 MB]

Operator: Wedge Product
Variant: New Form-0, Form-1, [0.053959 ms, 0.308864 MB]
Variant: New Form-0, Form-2, [0.257666 ms, 0.409744 MB]
Variant: New Form-1, Form-1, [0.154208 ms, 0.35056 MB]

[GeorgeR227]

Test Summary: | Pass  Total   Time
CUDA          |  139    139  41.7s
     Testing CombinatorialSpaces tests passed 

shell> git log -1
commit 67ad092ad6615e66994ce3956d7a65657fdf3684 (HEAD -> gr/clean-fast-dec, origin/gr/clean-fast-dec)
Author: GeorgeR227 <78235421+GeorgeR227@users.noreply.github.com>
Date:   Tue Jun 4 15:32:09 2024 -0400

    Added some doc strings.

[lukem12345]

Can you edit your comment to use the commit at which you ran those benchmarks

[lukem12345]

Checking that this is the dual vertex that is at the center of a primal vertex may just "fall out" of using our accessors in a certain order. So the following reliance on checking if it is in 1:nv(sd) may be unnecessary.

[lukem12345]

Further, this may be faster using @inbounds