Inline protocol parser for improved throughput

Replace the modular parse() function with chunk-based inline parsing in
_read_loop. Reading 64KB chunks and parsing multiple messages per buffer avoids
function call overhead on the hot path.

Handler signatures now accept bytes instead of str to defer decoding until
necessary.

Subscriber throughput improves ~63% (295k to 480k msg/sec).

Signed-off-by: Casper Beyer <[email protected]>

Author: caspervonb