Resistor feed is fine, provided that you have a good
voltage source tolerance, can stand the variation in
the "pilot" FET and so on. Simplicity often wins.
But if you -need- the current to be a particular,
accurate value (without re-twiddling the pot every
time somebody exhales) then a current source bias
might be preferable. Harder to come by than a resistor.
A higher voltage source, higher resistor value will
reduce the influence of make (process) and temperature
variations on the current taken by the pilot FET. You
can assume the FET will still hold itself down (limit it if
you feel shy) against a nominally-same current.
If you are concerned about wiper skip, noise while
adjusting, microphonics etc. you might put the pot
in shunt across a fixed R1 (or pot + series R1, in
parallel with R2) etc; you can get to a min and max
limit that remains sane with pot open / shorted.
I have myself used the resistor scheme similar to
shown, just for simplicity's sake. Other than, not
using the pin connected shunt cap (as this is an
AC short and for current-source-like you want an
AC open, infinite Z, or tolerable approximation).