It’s great that you have found and fixed the fault!
Reconciling the cause and symptoms is not so easy though. I have pondered it somewhat, and have come up with a possible explanation for why the disc input was apparently differently affected.
The tape adaptor board has a single emitter follower in each channel. This is configured to provide a 100k input impedance. If a collector connection is lost, then the incoming signal would see what is effectively a P-N (base-emitter) junction diode with a positive bias (around 6 volts) on the base. This would present only a small impedance to the incoming signal. And that diode would be in series with the emitter resistance, which totals 4k9. In turn that would be in parallel with the volume control (50k), and from the source viewpoint, also in parallel with the 100k emitter follower input resistance. (Two 220k bias resistors effectively in parallel as far as the signal is concerned, and a 1M bleed resistor ahead of the input capacitor). So the source would be looking at 4k3 give or take whatever adjustment is needed for actual diode resistance.
A line source with a buffered, low impedance output might not be too bothered by such a low impedance load. If the source impedance is 1k, then the loss as compared with a 100k load is 1.8 dB (neglecting any additional bass loss due to coupling capacitors sized for higher load impedances), probably only just noticeable.
In the case of the disc preamplifier though, the situation is different. Whilst the output impedance of the feedback pair is probably quite low in and of itself, it is looking into an R-C network that is designed to correct the HF error that is inherent with series feedback RIAA equalizing amplifiers. In this case there is a 3k3 resistor and a 1500 pF parallel capacitor, for a time constant of around 5 microseconds. The last-mentioned is situation-specific, and I should guess is calculated for the M2 disc position, where the unwanted HF turnover would occur at a lower frequency than with M1 position because of its lower gain, 25 dB as compared with 34 dB.
Anyway, looking back from the emitter follower, the disc preamplifier appears to have a 3k3 source impedance. When loaded by 4k3 as in the broken collector connection case, the signal attenuation will be around 5 dB, which would be quite noticeable. Not only that, but the time constant would be modified – I haven’t done the calculations – and it might be that there is some additional treble rolloff.
Cheers,