Optimizations with performance metrics #12

To be honest, the parser definitely doesn't perform badly. Using some "good enough" benchmarks, I have discovered that the ini file parser performs roughly the same as or better than python's built-in configparser for ini files up to 30KB in size on my ThinkPad. That's pretty big, and the obvious reason for the cutoff is that our parser performs a linear search across sections and keys on query, as opposed to a hashed dictionary data structure that python uses regardless of the size of the ini file. Our solution is simply more cache-friendly than python's on smaller files. Arguably, this is hardly an issue at all, and this issue is not very urgent for that reason.

To resolve this issue, I would like someone to look over the code and apply micro optimizations where they can fit without sacrificing simplicity and readability. If you do this, provide some performance metrics on a large ini file compared to those of a simple python script that parses the same file for the same values. If it's substantially faster under -O0 through -O2, then I would consider it a resolution to this issue and a big step towards v1.0.0