The block_buffer value in flush_block_buffer may be too aggressive

[telarium]

I was running into performance issues when writing Python code that would set a single block every few seconds, resulting in some fairly significant frame hitching.

Eventually I was able to track down the problem to the "vm:write_to_map()" operation in the flush_block_buffer. Currently it is writing to the map even if the block_buffer contains only a single block.

if #block_buffer >= 1 then

I would suggest changing this to a higher number so that it doesn't copying a 16x16x16 mapblock every time a small number of blocks have changed.

I'm running this project on a 32-bit ARM processor with OpenGLES v1 support.