PixelCNC v1.12a - curing memory allocation woes

v1.11a proved to be a vain attempt at resolving a memory allocation insufficiency and caused problems for some users whose systems could not find large enough contiguous blocks of unused memory to allocate just for opening up images. The fragmentation-free strategy of allocating a large block for each sub-system that's hopefully large enough for every situation PixelCNC may encounter exhibited the unacceptable potential for catastrophic failure when memory simply was unavailable.

v1.12a has been released which now properly allocates memory as-needed, instead of as large all-encompassing blocks, and frees it back up when it's no longer needed. This involved rewriting code wherever the allocated memory was used to store lists and tree data structures within which pointers to absolute memory locations elsewhere in the block were used. Pointers are now replaced with absolute indices, and the blocks serve as allocation pools, and are re-allocated larger if they cannot provide enough storage for data.