L2 is a pretty decent brick-wall limiter. I find the best way to use it is to use the link button between the sliders to pull both down until the sound starts to get rubbish. Then delink the sliders, set the output one to below 0, say -0.3dB and the one controlling compression so that only the peaks are getting hit, and then only by a very few dB.
I agree with Anderton about the L3, it’s very good and very flexible.
Cakewalk’s Concrete Limiter is/was very good but I don’t think it’s available now. Frontier does a different kind if job to a brick-wall limiter. L2 and similar limiters should be the last thing in the processing chain other than meters, Frontier I think is better used as a more moderate processor.
Compression before final limiting can often be a good idea, the Waves SSL bus compressor is good for that and is often on sale these days. The idea is to do things a little at a time - mastering is adding final touches, not making major changes to the audio. At least, it is unless you’re mastering for vinyl where the physical requirements of the cutting lathe, press and playback equipment might mean some significant eq is required to tame frequencies vinyl can’t handle well and there’s also the reduced dynamic range of vinyl to consider. Vinyl mastering is a really skilled and demanding job.
Don’t add any dither unless you are reducing the bit depth. If the project is, say, 24 bit and the exported audio file will be 24 bit switch off dithering. If you don’t the dithering will just be adding unecessary noise. It’s a process that really should only be carried out once as the last step in reducing the bit depth to that required by the target media. As for which dithering algorithm to use, if you can hear significant differences between them on anything other than a very simple, open mix congratulations, you have more sensitive ears than 99%+ of the population. :-)