A major reason that the NTSC (1st) chose negative modulation was because of the apparent ease with which black-level agc could be obtained simply by using peak sync amplitude as a proxy. Reality was different, and it was soon found that noise spikes, prevalent at VHF, particularly in Band I, upset both simple agc systems and simple sync separators. Thus gating quickly became the norm in all but the lowest priced receivers in the USA, and I imagine that the same was true in all those countries that used negative modulation systems.
For agc, line-gating using a line flyback pulse became the common method in the valve era. This could be done quite simply using one pentode or one triode, which method also provided quite a bit of agc gain. Or in other words it provided an amplified agc system that was probably no more complex than if one had attempted to amplify the agc voltage derived from rectifying the vision signal, whether on a mean or a peak basis. One way of looking at it is to think of the agc valve, say a pentode, as busily rectifying line flyback pulses and as a result providing a DC output from its anode. The magnitude of this DC voltage is determined by the valve current, and this is in turn determined by its grid bias as it happens to be at the time of conduction. Video signal is applied to the grid, but the valve conducts only during the very short line flyback pulse, which occurs either during the line sync pulse or during the back porch, and so is directly related to black level. The shortness of the conduction period means that noise pulses have no effect on the DC output unless they occur during the line flyback period. So that common single-valve agc circuit quite neatly did several jobs, and the line flyback pulses not only provided the timing but also the “energy” for development of the agc voltage, allowing DC gain that would otherwise have been not so easy to obtain with valves. Even if for some reason you decided that you didn’t need a line-gated agc system, it could be that the “standard” circuit was the lowest cost way to obtain the required gain. For example, the late 1950s Mullard sync-cancelled circuits used sync cancelling for timing, to access black level and for noise cancelling, but they still included line flyback pulse rectifier to obtain DC gain.
In about the same time frame, noise-gated sync separators were developed, typically using either dual-control pentodes or heptodes. In American practice, initially the 6BE6 radio mixer was used where a heptode was required, but then ad hoc short-grid base heptodes were developed, such as the 6BZ6 and 6CS6. In Europe I think that Philips started with the ECH81, then moved through the ECH83 to the purpose-designed short-grid base ECH84.
A late 1950s development in US practice was noise-gated as well as line-gated agc, for which the 6BU8 dual, dual-control pentode was introduced. One half did noise-gated sync separation, and the other half noise-gated and line-gated agc.
In the transistor era I think that there was some use of noise-gated or noise-cancelled sync-tip agc, without line-gating. With transistors it was easy enough to provide DC gain in the agc system, so gating could be separated from gain.
Early IC era practice was to provide for line-gated agc, as seen in say the MC1352, the TCA270 and the early jungle ICs. But then the Philips TDA2540/2541 of the early 1970s had noise-gated sync tip agc, with no provision for line-gating. This may have been to facilitate its use in VCRs, which did not have a readily available line flyback pulse. Other vision ICs of the period allowed the user a choice in agc implementation, line-gated or not, and if not, peak or mean level.
Overall, I’d say that negative modulation receivers covering the VHF channels mostly had line- or noise-gated agc, and where not, then they used the peak (sync-tip) type. Those with simple mean-level agc were probably few in number. My recollection is that even with line-gated agc, ignition interference of the kind that was quite prevalent back in the 1960s easily upset Band I reception.
The story may have been different with positive modulation systems. In Belgium, multistandard receivers were used, covering negative as well as positive systems, and the few valve-era schematics that I have seen show line-gated agc for both. It may have been different in France, where positive-only receivers were the norm. Perhaps a telling point is that the Philips TDA2542 vision IC, which was the positive modulation counterpart to the TDA2540/2541, had mean level agc, albeit with an external link into which I imagine that a gating circuit could be inserted. But not all of the IC makers did the same, and late in the analogue era were ICs that for positive modulation had agc systems that looked at white level as well as black level, and had internal gating generators that were timed by internally separated sync pulses.
In the UK I think that there was some interest in gated black-level agc systems (and black level retention or restoration) late in the 405-only era, but the arrival of dual-standard receivers seemed to be an excuse to abandon that. Somewhere I have seen it written that black-level agc on both systems would have been too complex and too costly to do in the valve era. But the Belgian setmakers managed to do it without undue complication. It was surely even easier with transistors, and I think that the Decca Professional comes to mind as having had quite a neat line-gated agc circuit for both 405 and 625.
I have seen only one or two schematics for valve-era British export sets for negative modulation systems; as I recall these had line-gated agc.