First, the amount of light you use is (close to) the minimum for the task - you're already doing far better than most. It may still be a concern as you seem to be in an area that's otherwise dark, but a good bike light is mounted fairly low and pointing down (equivalent to shading the light appropriately as mentioned in the Guardian article).
The article mentions that LEDs allow better wavelengths to be selected, but also say that blue-white lights are worse than others. A white LED is a blue LED with a yellowish broadband phosphor coating, so it's fairly strong in the blue region of the spectrum, which is less good. However to see where you're going and avoid hazards you need a broad spectrum. To test, and don't do this while riding, set up a view of something like a branch covered in green leaves, lying on a dark road, and illuminate that with your red rear light. You'll find it's much less visible with this narrowband source than with a white light. Switching to yellow or green just means other hazards disappear. You can get round this by using multiple wavelengths, but may well need more total light than with your well-designed white LED lamp; you'll certainly illuminate a bigger area. The journal paper and the citations in it state that several diverse species of insects are more sensitive blue and green light than to longer wavelengths. As you need these wavelengths to see well enough to ride, a wavelength-optimised source isn't a solution.
LED bike lights flicker (too fast for human perception) when run at reduced power (and sometimes at full power). This is because the power output is regulated by pulse-width modulation. Potential Biological and Ecological Effects of Flickering Artificial Light, Inger et al., 2014, cited in the paper linked in the question goes into some detail on flicker frequencies. Unfortunately it's a little old when considering the prevalence of LED sources, and concentrates on lower frequencies, however it's reassuring as figure 1B shows a few hundred Hz as the upper limit for insect sensitivity to flickering. A rough estimate of the flicker rate of my bike light is a few kHz, which is too fast to be an issue. This estimate is based on the streak length of water drops thrown off my front wheel, my speed, a lot of assumptions and mental arithmetic; I really ought to check with a photodiode, but this is a reasonable (or even low) value for LED PWM.
The methodology for assessing whether a given light is problematic is inherently a little complex, but as a rule of thumb, dimmer, redder, more specific to the need, and either steady (DC incandescent) or flickering at >1kHz are all good.
The paper also states that some modern LED fixtures emit ultrasonic frequencies that could have compounding effects on insect fitness. This would be due to switch-mode power supplies, used in most high-brightness LED lamps as they're more efficient than linear regulation. One small source passing by fairly quickly isn't going to be much compared to LED streetlights.
