The HF pulse needs to go into the workpiece and not back into
the source. So you will want a narrow pulse and a lot of source
inductance. I would check that your DC source indeed has that
kind of inductance.
You might prefer to deliver the HF to the torch directly so that
the welding cable inductance is in your favor rather than against.
Look to your flyback transistor turnoff time and snubber circuit
as ways to improve the output pulse.
An automotive ignition coil will likely produce a hotter spark,
more easily (perhaps too much kick, I've been knocked cold
by playing with them).
You do not need a lot of current in the HF, you only need to
ionize the gap so the real current can do the work. You want
maybe 8 zaps per sine wave cycle (if you're working with sine
wave AC power, which is not so popular anymore) or you could
sync the thing to the edges of the square-wave AC if it's an
inverter style deal because you only need to initiate when the
arc has snuffed at zero-crossings.
Your TIG working distance is only a couple of mm anyway so
what you're getting already might do (presuming it doesn't
get lost when hooked to the welder output).