use std::vector inside lcArray class

[ScanMountGoat]

This replaces the custom dynamic allocation logic for lcArray with std::vector. The standard library vector grows by some factor when resizing to make the insertion cost average out to be roughly constant. The old implementation for lcArray appeared to be resizing every N elements for some user configurable N. When loading a large model with roughly 58000 pieces, Project::load takes about 3000ms on my machine for the old lcArray and 2800ms for the new implementation with std::vector when profiling in Visual Studio. Application memory usage in task manager seems to be identical for both versions. Removing lcArray entirely is more involved, so I preserved most of the existing API.