After seeing one of my videos , which shows an american GE 150W HPS lamp in the parking lot of Castra mall cycling (GE because of the external amalgam reservoir), I had the idea, that these GE HPS lamps (Which during cycling are totally dark for 1:40-2:00 mins, than usually just restrike with a bluish MV colour , much like a penning start HPS lamp in a similar conditions), are Lucalox-I internal ignitor lamps, since they haven't a starting spiral, and for sure something inside the lamp prevent them from fire before 1:40 mins have been passed (And this can't be caused by a smart ignitor).
However, since in the case of the posted video, after 1:40 mins of a total darkness, instead of a light from a glow starter, an inner arctube glow that looks like the cold cathode dischage that made by a superimposed ignitor, and not from an external starting electrode, appears non-stop until the lamp hot restrike, I think that there is an option that these HPS lamps have an internal electronic ignitor with a thermal disconnector, which why they behave such.
It was possible that GE USA put also internal electronic ignitors with a thermal disconnectors in their Lucalox internal ignitor HPS lamps in the past?

I doubt there would be an electronic ignitor inside of the lamp, there are too high temperatures.
An electronic ignitor could be in the fixture (it is indeed not any programmed type, asthese have the "cool down" time ~3..5 minutes for HPS and that would mean direct restrike without the glow).

And couldn't the lamp be made by Iwasaki? Then it would be a FEC ignitor (Iwasaki patent) - that is integrated inside of the lamp (maybe with a thermal cutout) and generate similar pulses as an electronic starter (2 pulses per cycle, continuous pulsing till the lamp start)

But it could be some of the selfstarting types: The arctube is quite short and thick for an European/US S56 HPS, but it could be the S55 US type, which I won't be surprised if it start at 230V mains...

I and imj thought that this is a GE lamp, because of the external amalgam reservoir.
But since this can't be a penning start HPS lamp, since it haven't a starting spiral, so probably this is an Iwasaki FEC HPS lamp (Which their starter also disconnected when hot).

Update: Medved: I've seen the videos again, and the labels says S55 indeed and not S56 (I update the thread subject accordinally).
You said that you wouldn't be surprised if these lamps behave like this because they hot restrike from plain 230Vrms. But then they are very old and redded out, and hot restrike at bluish mercury color, and a lamp in such conditions, can't be hot restriked from plain 230Vrms anymore. So some sorts of electronic ignitor must be present inside the lamp to make it glow dimly like an external ignitor, and a bi-metal prevent it from operating after 1:45 have been passed, in the first above video.
These lamps are for sure GE and not Iwasaki, because of the presence of the external amalgam reservoir in them.

Update:
http://www.youtube.com/watch?v=qonxdBPBtFI
Is this video should answer my question about the strange cycling of the GE 150W S55, that I posed here, that these lamps indeed hot restrikes from plain 230V (Regardless of their state)?

S55 is designed to operate with lower arc voltage at higher current, so with 120V mains it suffice with simple series choke ballast.
But the lower arc voltage mean more power is used just to keep the electrodes hot, so these lamps have lower efficacy.
On 120V the use of this concept make sense, as the ballast is way cheaper and more efficient than an autotransformer based ballast, so the overall system efficacy become the same (150W) or even better (70W and below).

But on 230V the low arc voltage tubes do not make as much sense anymore, as beside the lower efficacy of the lamp itself, the lower arc voltage mean higher current in the ballast, so the ballast have to handle higher apparent power, so have higher losses. Therefore I'm a bit surprised to see the S55 used in a 230V mains environment.
Because it is so short, the 320V peak could be sufficient to ignite it even without any other pulser. But on 120V it have to be always used with an ignitor. And that mean it will behave differently on 120V vs 230V circuit.

The tiny glow just before it ignite is the corona discharge. It could happen, than the corona is not powerful enough to create a complete ionized path between the main electrodes, so only some small area around the electrodes glow there.
Usually this happen as a result of high dV/dt of the pulsing ignitors (electronic, FEC,...), where the steep edge ionise some gas around the electrode, but the voltage is insufficient for the discharge to spread too far.
But conditions for this not enough powerful corona could be created even by other reasons (e.g. because the antenna aid is still disconnected when the corona appear, so there is no aid to spread it through the complete tube, or the antenna is connected too early, so the corona appear way before the pressure allow the discharge to spread along the complete tube,...).

With shorter and/or wider arc tubes the corona usually always spread across the arctube, so create at least a spark between the electrodes, if it does not ignite the main arc (e.g. in MH).
It is the thin/long straight tubing shape, what help to create the conditions for only the corona around electrodes, without forming the complete discharge path. Therefore you could observe this phenomenon mainly on HPS and fluorescent.

Medved: In this video IMJ refering to the 50hz ballast operation frequency as a factor here for the lamp behavior during the cycling, and confuses the mind there. But this is indeed seems that this lamp starts directly from the ballast OCV (the ballast is an S55 autoregulator ballast), and that in videos that a glow at the upper part of the tube (What you calls a "corona discharge"(?)) just before the arctube restrike, fromed also from the ballast OCV, as there is no any starting aid inside the lamp, and the lamp seems to be just a plain external ignitor american GE Lucalox E 150W for S55 ballasts, that the ignitor inside failed, so the lamp starts directly from plain OCV.

Here the corona discharge is explained...
It is, what first happen around the electrodes when the lamp is getting the ignition voltage. To be ignited, the corona have to spread between both main electrodes, then it turn into a spark and eventually an arc.

I would guess the lamp was operated there on a simple series reactor ballast, I would not expect more fancy there...

The high frequency causes the corona discharge to operate at higher power, so become more potent to ignite the lamp (more ionization).
But it does not alter the distance, how far it reaches, for that you need the voltage.

And if the lamp does not have any starting aid, I would indeed expect the corona will first build up around the electrodes and extend it's range only as the pressure drop.
The starting antenna tend to "lead" the corona along, so the lamp need lower voltage for the corona from both sides to touch each other...