I'd say option two, because when you use I2C (SDA, SCL) between two chips, it's best practice to keep all connections as short as possible, and not to run the 2-wire signal through physical disconnects; sockets introduce problems, as do separate ground planes.
Even at the protocol's lowest speeds, keep conductive interconnects short to avoid reflection errors in the data stream (which I'm not sure the FV-1 can recover from gracefully as it loads programs) and to avoid putting high frequency noise near the op amps, which may cause gain overload (oscillation) at inaudible frequencies, and mess with the sound of the guitar.
I'd suggest trying it both ways but with an oscilloscope set to about 200uS on your audio signal to check for stray RF from the scl/sda pair, which you might twist several times to cancel reflections, unless you go with shielded cable.
I'm not being a know-it-all ... I'm humbly sharing what I've learned about mixing I2C and audio within a music project. I use arduino chips to control FV-1s in my personal projects, which use an sda/scl pair from a rotary encoder to choose which effect on my Arachnid board to engage, and another sda/scl pair to send that effect name to a display. I have a box of "mistakes" PCBs to back up the above suggestions about signal quality, data paths, and shielding with this signal combination.