Use ffmpeg (terminal tool).
Copy the audio into a separate file.
Then use ffmpeg again to silence desired part (https://stackoverflow.com/q/38180809/2792138 or https://stackoverflow.com/q/29215197/2792138).
Then merge new audio file with the initial video file, replacing ...
If it is Windows then why not to use dshow instead of gdigrab?
Please look at the
If both video and audio would be via dshow then you could use ffmpeg like this:
ffmpeg -f dshow -i video="VIDEO_DEVICE":audio="FIRST_AUDIO_DEVICE" -f dshow -i audio="...
You need to use cable specifically designed for SDI. RG59 essentially only defines the impedance and capacitance of the cable, not the return loss and attenuation characteristics. You can use reference charts from cable manufacturers like Belden and Gepco to select the right cable for your situation.
When you record each track, make sure that the level never goes up to zero on the meter. If it hits zero, it means that the recording is clipping. Clipping is when the computer can no longer accurately record the waveform. So instead of recording the shape of the wave, the upper and lower limits of it will be clipped in a straight line. So a tone sound ...
Save the money for the on-camera mic, it is by far the worst option. Other options include:
lapel or lavalier mics. Wireless lav mics are as you say expensive, but wired lav mics might be the cheapest option. You'll need something that can take 3 inputs, either a multitrack recorder, or a simple mixer to mix it to stereo (I'd mix the two subjects into one ...