This is a super common issue and it's linked to the power supplies. LED street lights use HF switchmode power supplies with feedback loops to their controllers. All it takes is one component to fail on the feedback line and you have a flashing lantern. This is the biggest reason why LED is total (...) for outdoor lighting and should be limited to municipal and indoor.
The offending failure is most likely not even inside of the ballasts, but in the installation wiring.
The difference towards classical magnetic ballasts is, the magnetic ballasts tend to increase the input current when the input voltage increases. In case it does not mean higher load power, nor higher losses, it causes higher reactive power. But the important tging is, the current increases. This inherent feature means, if some fault makes the system effectively running lanterns in series, it balances itself around some working stable equilibrium operating point, so everything still works at least somehow (even when the performance degrades, usually it is stable). The same is valid for many electronic ballasts for e.g. fluorescents and/or induction (when the ballast does not feature any constant power regulation)
The electronic ballasts on the other hand strictly maintain the output power and the input as high power factor resistor. That means if the input voltage rises, the input current drops. So it exhibits a form of a negative dynamic resistance in its characteristic.
The negative resistance means, the eventual equilibrium point does exist as well, but contrary to magnetic ballasts, with electronic it is unstable. That means volage across one ballast rises, across the other falls. The first then reduces its current, the other tries to increase it. But as the series combination always yield the same current (1'st Kirchhoff's law), the voltage across the second is dropping faster and faster, till it reaches the undervoltage shut down, so does the connected LED.
At that moment the ballast stops regulating, becomes rather high impedance on its input, so even the other ballast in the series string looses power and shuts down.
Then the voltage across it rises, till the ballast restarts, so again you get an unstable condition with two ballasts, so the cycle repeats, hence the flashing.
The explanation above was made on a simple example of just two ballast in series. In real installation (with many ballasts), with a real life fault (e.g. loose Neutral in multiphase supply system) the network is way more complex, but the instability is still there. Just the high amount of ballasts leads to high amount of state combinations they may achieve, so the flashing is no longer as regular as with the simple example.
