After correcting the timing RC network and the IR receiver power supply filter, I'm happy with the performance of the modules, at least judging by the transmission and reception of 1 kHz test signal. The pulse frequency is near the target 57.6 kHz and the signal from the receiver is sharp and high.
Right after the assembly, a basic test confirms the modules work, though the pulse generator frequency is somewhat off. Replacing a 6.8 kΩ resistor with a 4.7 kΩ should fix this.
When one module is fed with the 1 kHz signal (from the oscilloscope reference generator), the other receives the signal with a slight delay (as expected).
Funny that the practical formula for the pulse generator frequency is f = 1 / (3.14 * R * C), not f = 1 / (2.197 * R * C), as advertised in the source. Some early π-day interference here!