how excactly is the high voltage pulse created in a lamp ignitor?

Never studied lamp ignitors, but from what I've heard, it essentially shorts the ballast output to lamp for very brief pulses creating an inductive kick many times a second. I'm sure there is more to it though....

i cracked open the HPS Ignitor of a Lithonia Yardblaster and found a bobin with three small coils of fine wire,a 5 watt ceramic resistor,a 270 volt cap,and some small black component.

I already have some exposed ignitors that has parts on PCB board....

THe basic 2 wire ignitor essentialy shorts the ballast momentarily at or near the peak of the mains voltage, and then interrupts the short near the end of the cycle. This creates a high voltage pulse to strike the lamp.

The 3 wire igniter has a coil inside, that is in series with the supply. They superimpose a high frequency switching action on the output to provide the high voltage using this coil.

Certain ignitors that use a tap on the ballast basically turn the ballast into an autotransformer to provide the high voltage, without generating RF noise, and being able to start a lamp a distance away from the ballast as long as the lamp wiring is rated to withstand the full starting voltage ( around 2kV). I have only seen these in 220V versions, not for 110V.

Basic ignitors will repeatedly try to start the lamp at EOL or if it is removed, eventually failing either open circuit or shorted. The more sophisticated electronic types will try for a short time then shut down until power is cycled.

It would be nice to find a schematic of the three wire version to repair the older style PCB versions used by GE on M-1000 HPS.

Depends what ignitor we are talking about.
The two wire ones may in theory work like Sean explained, but that is the case only forfluorescent preheat starters.
Otherwise the circuit use to be a bit more complex:
For LPS it is form of a HF oscillator (capacitor in series with a switch element, plus control part).
Some (high power MH) use a kind of Delon multiplier separated by a resistor, plus a kind of spark gap discharge element. The multiplier generates Varc*Constant and charges a capacitor with it. When the lamp has no arc, the ballast OCV makes that capacitor to charge above the breakdown voltage of the spark gap, so it connect the high voltage from the charged capacitor(s) onto the terminals, forming a pulse there. Once the main arc is ignited, the voltage is too low to reach the breakdown on the spark gap, so it is inactive.

The three wire ignitors are of two types according to how they work and one of them has two types on how they are implemented.
The one described by Sean is the superimposed type, so it contains all HV generating components. Its advantage is it does not expose the ballast winding to the HV (nor the connection between the lamp and ballast so it could be long, so remote ballast with ignitor close to the lamp), but disadvantage is the secondary of the pulse transformer is in the main arc current path, so mainly with high power lamps it needs to be beefy.
Variant of the same functionality is the semiparallel circuit, where the ignitor uses the ballast coil as the HV pulse transformer: The ballast has a tap few turns from the lamp end. The ignitor then discharges the capacitor into the few turn section of the ballast coil. That induces the HV pulse to the rest of the ballast winding, so the HV starting pulse for the lamp. These are common in the US, as it sufices with just 4 small components. But exposes the ballast winding to the high voltage.
The last major topology is again using the resonance between the capacitor and part of the ballast coil (~20% from the mains side on series choke), with a switch (triac) in the circuit generating the high frequency oscillations. The oscillator uses to generate up to 700V peaks, the ballast coil acting like a pulse transformer, multiplies that about 5x, so forming the 3.5kV or so pulses from it. Common with series choke ballasts on 230V (e.g. Philips SN57). Advantage is, it can handle quite long wire to the lamp (ir includes it into the resonance equation), so may easily act as a remote ballast/ignitor, the drawbacks are exposing the ballast and wiring to the HV, plus passing the HF primary currents via mains, so needs the PFC capacitor to be present and wired close together with the ballast and ignitor.