I’ve seen lots of people end up with transistors in hand that had a slightly different pinout than what the guy who designed the board thought it would be.
They usually get 2 legs soldered down, then run a very short wire-wrap wire from the last leg to the other pad.
Or were you lucky enough to get pFETs that dropped right in?

I see that, like me, you turn on the power and program the CPU long before all the parts are soldered down.

As for rows not turning on I have realized that the problem is that I need a smaller resistor since current controlled devices vary the voltage to get the specified current. I still am not sure if running this short circuit will have an adverse effect, however, so it is still an option.

The actual power supply hasn’t been designed yet, so at the moment my bench supply will have to suffice until my power draw gets over 800mA.

Would it be too much work to both put a surface-mount footprint for the small crystal on your PCB, and also extend the traces to the large crystal?

Then when you get the PCB, you can start putting on the small crystal first, and then either
(a) scream in frustration and throw the impossible-to-solder SMT part across the room, then smile as you fit in the through-hole part, or
(b) smile as the SMT part solders right into place, and you’re left wondering why people make such a big deal out of SMT. Then (optional) saw through the unnecessarily-long traces that lead to the now-unused through-hole footprint.

Either way you are left smiling.

I am almost certain that driving the gate to a mere 0.26 V above +5.0 V is not turning your high-side nFETs completely on.
Most (all?) nFETs need Vgs_on over 3 V to completely turn on.
The other end of the nFET (the end we are *trying* to pull up to turn on the LEDs) is pulled up to, at most, the Vgs_threshold voltage below the gate voltage.
(oversimplifying:) If your Vgs_threshold is 3 V, and your gate voltage is 5.26 V, then the output is a mere 2.26 V above ground — perhaps that’s why shorting out the column driver resistor seemed to help.

I don’t like running digital logic chips off nonstandard power voltage.
But if we give the row driver digital logic its standard power voltage, how do we turn the MOSFETs completely on?

There are at least 2 ways:
(a) Replace all the hi-side nFET with hi-side pFET. Turning on high-side pFETs by pulling the gate a few volts *down* from +5V, is much simpler than turning on nFETs by pushing the gate a few volts *up* above +5V.
Or,
(b) Use some sort of “level translator” “high-side driver” between the row driver digital logic and the high-side nFETs.
The level translator converts the 0 V to 5 V output of the row driver
to the ~1 V or less to turn the nFET hard off, and a “Vpp” high voltage of “5V + Vgs_on” or more (8 V or more?) to turn that nFET fully on.

The simplest level translator is a one-transistor NOT gate (with a cheap nFET or npn), with its pull-up resistor to the “Vpp” high voltage.

I hope you can easily get the Vpp you need (8 V? 10 V?) from your power supply.
There are many other clever ways to generate (“bootstrap”) that “Vpp” voltage.