Fix packet compression handling

[ItsDrike]

This fixes a bug in how we previously handled packet compression. Originally, our handling was done based on a boolean, either do enable compression, or don't.

However this doesn't really work, as the way we recognize whether to enable compression is through the server, sending us the LoginSetCompression packet, which only contains a compression threshold value. This value can be -1, in which case compression should be disabled, it can also be 0, in which case compression would indeed always be enabled, but for any other value, compression should only be enabled if the packet length surpasses this threshold, and we only know the packet length during writing, so our interaction functions need to be aware of it.

Note that if this threshold isn't crossed, but is set, the packet format still changes (there is an extra varint after packet length, which generally contains the uncompressed length, but when compression wasn't used - below threshold, it will be set to 0, and the packet id and data fields below remain uncompressed).

This PR changes sync_write_packet and async_write_packet functions and replaces compressed bool parameter to compression_threshold parameter, which is now being used properly. This however breaks compatibility, as these writer functions are a part of the public API, and the compressed parameter will no longer be available. Considering it didn't really even work properly before, it's unlikely that someone was actually using it yet, nevertheless it is a breaking change.

This does also affect how packet reading works, however there is no breaking change, as only the logic needed to change, to account for the possibility of the data length being 0, in which case compression wasn't used. This function still only takes a compressed bool flag parameter, rather than a compression_threshold, as if the threshold is non-negative, the packet format fundamentally changes, so we should already be reading the packets differently. For that reason, a bool flag is sufficient here, and should just be set to True whenever the threshold is >= 0.

Edit: Even though that makes logical sense, I've decided against making the reader use compressed bool flag, as it might be annoying for the user to have to do compressed=compresssion_threshold >= 0 every time when calling the function, as opposed to just compression_threshold=compression_threshold. So while the reader doesn't need to know this value, it also doesn't hurt to know it, and we just figure out whether or not compression is enabled ourselves, instead of bothering the user with it. Making this a breaking change too. The underlying _deserialize_packet function still only takes a compressed bool flag though.