Simplify frame length packing in TCP write

[jakirkham]

Noticed that TCP write was taking a while in the process of other benchmarking work. These struct.pack calls stuck out. Combined these into a single struct.pack call for simplicity. This results in a nearly 5x speed improvement.

Here's a simple benchmark to illustrate the effect of this change:

In [1]: import struct

In [2]: %%timeit lengths = list(range(10))
   ...: 
   ...: length_bytes = [struct.pack("Q", len(lengths))] + [
   ...:     struct.pack("Q", x) for x in lengths
   ...: ]
   ...: length_bytes = b"".join(length_bytes)
   ...: 
   ...: 
3.45 µs ± 88.6 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

In [3]: %%timeit lengths = list(range(10))
   ...: 
   ...: nframes = len(lengths)
   ...: length_bytes = struct.pack(f"Q{nframes}Q", nframes, *lengths)
   ...: 
   ...: 
711 ns ± 2.05 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)