Refactor c_buffer to make it suitable to re-use for a stack

[dylanahsmith]

Extracted from #59

• I changed the c_buffer struct to store pointers for the end of the data and capacity, instead of storing the sizes and capacity directly. This is because we are mostly getting and updating the data end pointer, which is especially true in order to use this for the VM stack where we reserve space for the block and don't have to check the capacity for each write/push.
• I've also added support to c_buffer to be initialized with size 0 (without an allocation) which is the one case where we can't just double the capacity until we have the requested extra capacity.

[peterzhu2118]

One minor comment, otherwise looks good!