For a variable lab bench supply you want one that has adjustable voltage that once set to a particular selection can supply a whole range of current to the load whilst the voltage stays constant.
Your first idea of using the pot between the +12 to -12V will not do do that. Pull any amount of current from the wiper of the pot and the voltage will change.
The second idea is a hack job modification done to the a switch mode power supply. From the description that you snipped this relates to a very specific model range of power supply and trying to apply this in general to any PC power supply from any vendor is not going to work. Not all power supplies use the same circuit configuration or even the same control chips. To be able to modify the behavior of any given switch mode power supply requires that you have good knowledge of how the circuitry of that supply is designed and how it works.
There is another reason not to be screwing with trying to make a switch mode power supply adjustable. Most switch mode controllers, especially the lower cost designs supporting high current outputs, are design optimized to vary the PWM of the switching output over a range to satisfy the varying current load range for the output. If you try to modify the feedback network to make the switcher produce other output voltages the design optimizations will no longer apply. As a result the supply may exhibit a number of undesirable behaviors including:
- The output may not adjust over a very wide range.
- The output may not be able to supply the full range of current at many output voltage settings.
- The switching regulator could become unstable and oscillate like crazy. This behavior could vary depending upon where the adjustment pot is set.
- The ripple of the output could be extremely high.
There could be other problems as well. Personally I think you would be better off purchasing a power supply specifically made for lab bench use. There are variable supplies available online that are fairly reasonable in cost. With these you also get the ability to vary the current that can be delivered to the load. This is a very useful thing when powering up a new circuit because you can set the supply up to protect the circuit in case there is an error in the circuit that would cause excess current flow that could potentially blow out some components.
As you wrote, this is a decoupling capacitor. When the current demand of a device increases suddenly, the power supply circuit may not be able to fulfil it immediately, e.g. due to the inductance of the traces.
Decoupling capacitors just jump in, until the higher current is delivered by the supply.
But real capacitors are not ideal capacitors. They also have a resistance and inductance. This may cause similar problems as for the supply. As larger capacitors have a higher inductance, the solution is to use a large and a small capacitor. The smaller reacts faster, the larger endures longer. 100nF and 10nF are very typical values.
But it's not absolutely necessary that you solder a new cap of exactly that value, 8nF, 22nF or 33nF would do the job, too, (also remember, many caps have a precision of just 10-20%). Also, it's possible that you don't need that cap to run the Pi.
If you only have 1mm solder, just put some (not too much) solder on the pads of the PCB, add some flux, hold the cap onto the pads with tweezers, and solder it with a fine tip, without additional solder.
If the cap is tiny, there is a chance that you produced a short circuit. You should test the resistance, it should not be almost zero. It's also possible that there is no electrical connection, but as said, it's possible you will not notice that.
But also note that soldering is thermal and mechanical stress for tiny components, especially when removing them. This may also damage your robbed caps, I woud always recommend to use fresh components.
Best Answer
According to the service manual, the potentiometer is connected between the signals called +5DA and AVSS:
simulate this circuit – Schematic created using CircuitLab
Replacing the 10k potentiometer with a 5k just means that it will draw a constant 0.5 mA more current, which will not be any problem what so ever. +5DA is connected straight through to the main VDD, so it can definitely handle it.