Does it support incremental parsing? For example, when I am parsing a network protocol, can it still consume some data from the head of the buffer even if the data is incomplete? This would not only avoid multiple attempts to restart parsing from the beginning but also prevent the buffer from growing excessively.