When small chunks are received from a large payload, i.e > 10MBs, performance
gets impacted because due to the buffer allocation, for payloads of 10MBs with
chunks of 2KBs it can take ~  20 seconds.

We add an exponencial growth of the buffer providing enough room or subsequent
chunks until a maximum threshold of 1MB.

With that change we can see a x50 improvement with the trade off of having to
allocate a little extra space, worst case scenario 1MB.

Beyond that number there is no gain in the performance