convertSubChunk* accept std::function(uint8_t, uint8_t) which puts an unnecessary performance burden on the converter

[dktapps]

The conversion has to merge the id/meta into a single number in order to find unique entries. However, this results in it having to take the number apart again in order to pass it to the mapper, which is currently always flattenData, which promptly proceeds to just put it straight back together again. Since the mapper isn't currently inlined, the compiler can't optimise away this useless crap which adds extra overhead.

https://github.com/pmmp/ext-chunkutils2/blob/master/lib/SubChunkConverter.h#L45-L46
https://github.com/pmmp/ext-chunkutils2/blob/master/src/PhpSubChunkConverter.cpp#L14

[dktapps]

As of 0cad64e, these are now able to be inlined, so the performance hit is reduced.