[Solver] Optimize Disassembler

[mbrown1413]

Currently the disassembler is fairly slow. Ideally, we rewrite it such that any Grid operations aren't on the critical path, similar to how the assembler already works. The current disassembler will still be necessary for custom arbitrary constraints in the future.

How GENDA disassembler works

One example of a disassembly that's slower than it probably should be: Slow Disassembly Test.json

Some other low-hanging fruit:

• Pieces are copied with all of their extra unneeded info, like label, bounds, etc. The disassembler doesn't need any info except the id, voxels, and attributes. Implement a method like Piece.clearExtraInfo() for working piece copies in the disassembler internals.