I agree that running status, by itself, isn't causing the problem. But if the MIDI interface is poor quality, maybe it is getting confused by the running status.
Here's a hypothesis I had, but there's not an easy way to confirm this hypothesis. If the MIDI interface is poor quality, maybe it is re-arranging the order of the MIDI messages that happen at the same time.
Hypothetical example
Keyboard sends:
(message 1) 90 3C 64 (Note On channel 1, middle C, velocity 100)
(message 2) 91 3C 64 (Note On channel 2, middle C, velocity 100)
(pause)
(message 3) 3C 00 ( " " " ", middle C, velocity 0)
(message 4) 90 3C 00 (Note On channel 1, middle C, velocity 0)
Interface: rearranges message 3 and 4.
PC gets:
(message 1) 90 3C 64 (Note On channel 1, middle C, velocity 100)
(message 2) 91 3C 64 (Note On channel 2, middle C, velocity 100)
(pause)
(message 4) 90 3C 00 (Note On channel 1, middle C, velocity 0)
(message 3) 3C 00 ( " " " ", middle C, velocity 0)
In this hypothetical example, message 3 uses running status and is supposed to turn off the channel 2 note. But if message 3 gets rearranged to come last, it ends up turning off the channel 1 note again, and the channel 2 note is left hanging.
MIDI-OX can't see incoming running status
If you search online for "MIDI-OX running status" you will find forum posts that explain Windows MIDI drivers add back the missing status bytes implied when running status is used before the application sees the message. (For example:
Running status indication (MIDI-OX User Forum)) Since MIDI-OX always shows a status byte, we can't troubleshoot if the original message from the keyboard was using running status by using MIDI-OX.
Continuing the hypothetical example above...
Windows MIDI driver: adds in status bytes implied by
running status.
MIDI-OX sees:
(message 1) 90 3C 64 (Note On channel 1, middle C, velocity 100)
(message 2) 91 3C 64 (Note On channel 2, middle C, velocity 100)
(pause)
(message 4) 90 3C 00 (Note On channel 1, middle C, velocity 0)
(message 3) 90 3C 00 (Note On channel 1, middle C, velocity 0)
Note: MIDI-OX calls a Note On message with velocity 0 a Note Off because it behaves like a Note Off. But in Jomarx Soares's video you can see they are 9_ __ 00 messages (Note On channel _, __, velocity 0).
My hypothesis is just a guess, but you can begin to see how a poor quality MIDI interface might cause problems like this. If some MIDI messages are re-arranged (or even dropped), then the running status might no longer be correct.