Largely this is because it's an inefficient proof-of-concept implementation, but in my experience it takes a lot of work to convert such a thing into a performant receiver.
I'm basing my stuff on SQ5BPF's work, who added the (purposefully) missing bits in osmocom's osmo-tetra project: https://github.com/sq5bpf
It works, but the code is not pretty and it's a really long pipeline of shell pipes and FIFOs. I'm working on simplifying the whole thing, I've already written a minimal replacement for telive. Hopefully I can change the pipeline from:
GRC tuner flowgraph -> simdemod2.py (pi/4 CQPSK demodulator) -> float_to_bits (pi/4 float to byte-per-bit stream) -> tetra-rx -> telive
to:
tuner.py (combined tuner, pi/4 CQPSK demodulator and pi/4 float to byte-per-bit stream) -> tetra-rx -> tetrarec.py (telive replacement)